AI skills needs and gaps for MSMEs in the retail sector in Cyprus, Germany, Italy, Poland, and Romania

AI skills needs and gaps o
MSMEs in he e ail sec o
in Cyp us, Ge many, I aly,
Poland, and Romania
AI skills needs and gaps o MSMEs in he e ail sec o
in Cyp us, Ge many, I aly, Poland, and Romania
Repo p epa ed by esea che s o DELab, Uni e si y
o Wa saw:
We onika Łebkowska: p epa a ion and analysis o
s a is ical da a
Michał Paliński: sys ema ic li e a u e e iew and ex
mining
Ka a zyna Śledziewska: gene al esea ch design and
o e iew, analysis o s a is ical da a, p epa a ion o
ecommenda ions
Ba osz Ślosa ski: esea ch design and o e iew,
quali a i e li e a u e e iew, hema ic analysis, ex
p epa a ion and modifica ion
Ka ol Teodo owicz: p epa a ion o use cases,
quali a i e li e a u e e iew, o e all esea ch
assis ance, ex p epa a ion
Rena a Włoch: gene al esea ch design and o e iew,
s uc u e o he epo , ex p epa a ion and
modifica ion, p epa a ion o ecommenda ions
Resea ch suppo :
Aga a Komendan -B odowska: design and ca ying
ou o co-c ea ion wo kshops
Resea ch o ganiza ion:
Michał Szos ek: o e all esea ch suppo
This publica ion ecei ed unding om he
Eu opean Union'’s Ho izon Eu ope Resea ch and
Inno a ion p og amme - G an Ag eemen No.
101133847. Views and opinions exp essed a e hose
o he au ho s only and do no necessa ily eflec
hose o he Eu opean Union o he Eu opean Heal h
and Digi al Execu i e Agency (HADEA). Nei he he
Eu opean Union no he g an ing au ho i y can be
held esponsible o hem.
The euse o his documen is au ho ised unde he
C ea i e Commons A ibu ion-Noncomme cial-Sha ealike
4.0 In e na ional (CC BY-NC-SA 4.0) licence. This license
enables euse s o dis ibu e, emix, adap , and
build upon he ma e ial in any medium o o ma o
noncomme cial pu poses only, and only so long as
a ibu ion is gi en o he c ea o . I euse s emix,
adap , o build upon he ma e ial, hey mus license
he modified ma e ial unde iden ical e ms.
How o ci e his epo :
Włoch, R., Ślosa ski, B., Paliński, M., Śledziewska, K.,
Teodo owicz, K, Łebkowska, W., (2025). AI Skills
Needs and Gaps in he Re ail Sec o in Cyp us,
Ge many, I aly, Poland, and Romania. Ve sion 2.
Zenodo. DOI: 10.5281/zenodo.12793437
Ve sion 2.0 (Oc obe 2025). Included execu i e summa ies
ailo ed o specific e ail audiences and efined syn hesis
connec ing esea ch findings o AI lea ning pa hways.
1 | 249
TABLE OF CONTENTS
How o Read This Repo – and Wha o Take om I 5
Execu i e Summa y o F on line Re ail Wo ke s 7
Execu i e Summa y o Re ail Manage s 10
Execu i e Summa y o Re ail S akeholde s 14
Execu i e Summa y o Voca ional Teache s and S uden s in he Re ail Sec o 18
In oduc ion 21
Aim o he Repo 21
Main esul s 22
O e iew o he Repo 23
Resea ch me hodology 26
1.The Po en ial o AI echnologies use in e ail 31
Key akeaways 31
1.1. AI uses in e ail based on Sys ema ic Li e a u e Re iew 31
1.2. AI uses in e ail based on quali a i e li e a u e e iew 39
1.3. AI unc ions in e ail acco ding o he e ail sec o s akeholde s 59
2. The s a e o AI adop ion in e ail sec o 69
Key akeaways 69
2.1. The main cha ac e is ics o he e ail sec o in he selec ed coun ies 69
2.2. The le el o digi iliza ion in he e ail sec o 75
2.3. Use o ad anced digi al echnologies 80
2.4. The le el o digi al skills 93
2.5. Cha ac e is ics o selec ed s a es 103
3. Real-li e AI uses in e ail companies: use cases and bes p ac ices 105
Key akeaways 105
3.1. ALOHAS 106
3.2. Feed. 108
3.3. G ünewald 113
3.4. Ke igans 115
3.5. Kuchyne Valen 118
3.6. L Cosme ics 121
3.7. Pe Media G oup 124
3.8. P ocosme 127
3.9. Pu elei 130
3.10. Repea 133
3.11. Sinne up 137
3.12. So mund 140
3.13. Velasca 144
3.14. Bes p ac ices - Inno a i e Re ail Labo a o y 147
2 | 249
3.15. Bes p ac ices - Foo p in s AI & Danubius 150
3.16. Bes p ac ices - Da e Media 151
3.17. Towa ds classifica ion o models o AI adop ion and in eg a ion in e ail companies 153
4. AI Skills: Defini ions and Classifica ions 158
Key akeaways 158
4.1. Impo ance o AI Skills 158
4.2. AI Skills Defined 159
4.3. AI Skills Classifica ions 161
4.4. AI Skills in Re ail: Sys ema ic Li e a u e Re iew App oach 165
5.AI skills needs: pe spec i e o employe s 173
Key akeaways 173
5.1. Da a collec ion 173
5.2. Tex mining analysis 174
5.3. Main insigh s 176
6. AI skills needs and gaps: pe spec i e o he e ail sec o s akeholde s 193
Key akeaways 193
6.1. AI Skills in Re ail Companies 193
6.2. AI Skills o Manage s 196
6.3. AI Skills o Wo ke s 200
7. Challenges o AI Adop ion in Re ail 205
Key akeaways 205
7.1. Challenges iden ified in he academic li e a u e 205
7.2. Challenges iden ified by e ail sec o s akeholde s 209
8. Recommenda ions and guidelines o designing an AI Co e cu iculum o MSMEs in Re ail 217
8.1. P ac ical guidelines o cou se design 217
8.1. De elopmen o ans e sal compe encies 218
8.2. De elopmen o echnical skills 220
8.3. F om Resea ch Insigh s o Lea ning Pa hways o AI Readiness 223
Re e ences 227
Annex 234
Appendix 1. Snapsho s om job po als included in he job ad e isemen s analysis (sec ion 5) 234
Appendix 2. Co-c ea ion wo kshop scena io 237
Appendix 3. In e iews and ques ionnai es scena io 240
Appendix 4. Sugges ed le els o skills and compe ences de elopmen h ough INAiR open educa ion esou ces as
p esen ed in he p ojec applica ion 242
Appendix 5. The scope o he Wo k Package 3 244
Appendix 6. Indica o s o he goal ealiza ion in WP3 246
Appendix 7. The lis o he s akeholde s who con ibu ed o he co-c ea ion o insigh s p esen ed in he epo 248
3 | 249
HOW TO READ THIS REPORT – AND WHAT TO TAKE FROM IT
The e ail sec o is complex and di e se. I b ings oge he company owne s and manage s,
ade o ganisa ions and unions, on line employees, and oca ional eache s and s uden s. All
o hem a e pa o he same ans o ma ion, ye each expe iences i di e en ly. A ificial
in elligence is en e ing e ail no as a single echnological e olu ion, bu as a g adual, une en
p ocess ha influences s a egy, daily wo k, and lea ning ac oss he sec o .
This epo cap u es ha complexi y. I is based on ex ensi e esea ch conduc ed in Cyp us,
Ge many, I aly, Poland, and Romania, combining li e a u e e iew, ex mining, job ma ke
analysis, s akeholde wo kshops, and case s udies. Toge he , hese da a p o ide a
mul i-laye ed iew o how AI is pe cei ed, implemen ed, and nego ia ed in he e ail sec o ,
highligh ing bo h oppo uni ies and ensions ha accompany echnological change.
The epo can be ead in mul iple ways. I o e s e idence, in e p e a ion, and o ien a ion o
di e en g oups wi hin he e ail ecosys em. I s pu pose is no o p esc ibe solu ions bu o
p o ide a cohe en , e idence-based unde s anding o he sec o ’s cu en posi ion –
knowledge ha has so a been dispe sed ac oss sepa a e s udies, indus ies, and na ional
con ex s.
Fo eade s om di e en backg ounds, he same ma e ial may o e di e en insigh s:
● Manage s may ocus on how AI a ec s business ope a ions and decision-making, and
on wha compe ences hei eams will need o manage echnology e ec i ely.
● Re ail s akeholde s – such as chambe s, associa ions, and ade unions – may find in
he da a a sys emic iew o he sec o and e idence o guide policy coo dina ion,
aining ini ia i es, o collec i e lea ning.
● F on line wo ke s may ecognise e e yday eali ies eflec ed in he findings – he
changes in asks, ools, and expec a ions ha accompany au oma ion.
● Voca ional eache s and s uden s can d aw om he epo an unde s anding o how
e ail wo k is e ol ing and how educa ion can p epa e o ha u u e.
These ou pe spec i es a e also eflec ed in he ou execu i e summa ies ha accompany
his epo . Each highligh s a di e en way o eading he findings: om business s a egy and
policy design o wo kplace expe ience and educa ion. They a e no sepa a e in e p e a ions bu
complemen a y lenses, emphasising ha digi al ans o ma ion is a sha ed p ocess in which
e e y ac o has a dis inc bu in e connec ed ole.
A a deepe le el, he epo iden ifies ecu ing a eas o compe ence ha a e c ucial o
na iga ing AI adop ion ac oss he sec o . These can be unde s ood as h ee in e linked le els
o lea ning and eadiness:
● Unde s anding and awa eness – building basic knowledge o wha AI is, how i unc ions
4 | 249

in e ail, and wha asks o decisions i suppo s.
● Applica ion and in eg a ion – lea ning how o use AI and da a-d i en ools in p ac ice,
in e p e hei esul s, and combine hem wi h human expe ience and judgemen .
● S a egy and eflec ion – linking AI adop ion wi h b oade o ganisa ional goals,
inno a ion, e hics, and sus ainabili y.
Taken oge he , hese insigh s o m a s uc u ed knowledge base ha p o ides a ounda ion
o u u e lea ning pa hways. The epo does no ye desc ibe hese lea ning p ocesses in
de ail, bu i poin s clea ly o whe e hey a e mos needed and wha hey should encompass.
In his sense, he epo fills an impo an gap: i b ings oge he a comp ehensi e and
compa a i e unde s anding o how AI ans o ms e ail wo k, knowledge, and o ganisa ion
ac oss mul iple Eu opean con ex s. This unde s anding se es as he g oundwo k o
subsequen phases o he p ojec – whe e esea ch findings e ol e in o p ac ical, accessible,
and inclusi e educa ional ma e ials designed o help he sec o lea n, adap , and g ow
esponsibly in he age o AI.
5 | 249
EXECUTIVE SUMMARY FOR
FRONTLINE RETAIL WORKERS
Sales s a , cus ome se ice, wa ehouse, logis ics, and ope a ions wo ke s.
WHY THIS REPORT MATTERS FOR YOU
The wo ld o e ail is changing. New
echnologies – especially a ificial
in elligence (AI) – a e becoming pa o
e e yday wo k: om cha bo s and
ecommenda ion sys ems o sma shel es,
wa ehouse scanne s, and online cus ome
suppo .
Fo many wo ke s, hese changes can seem
wo ying. Will AI eplace people? Will
machines decide wha we do? Ou esea ch
shows some hing di e en : AI in e ail is
no eplacing people – i is changing how
people wo k.
The goal o he INAiR p ojec is o make
su e his change is ai , inclusi e, and
beneficial o e e yone. We s udied wha ’s
eally happening in e ail companies ac oss
Cyp us, Ge many, I aly, Poland, and
Romania – how AI is used, wha skills a e
needed, and how wo ke s expe ience he
ans o ma ion.
WHAT WE LEARNED
AI is al eady he e – and i ’s g owing as
AI is no only o big ech companies. In
e ail, i ’s used o:
● answe cus ome ques ions online
h ough cha bo s,
● ecommend p oduc s based on
cus ome p e e ences,
● p edic which goods will sell bes ,
● help plan s ock le els and deli e ies,
● au oma e simple da a o epo ing asks.
These ools make wo k as e and educe
epe i i e asks. Bu hey also change how
oles a e o ganized: people spend less ime
on ou ine ac i i ies and mo e ime on
communica ion, eamwo k, and
decision-making.
Wo ke s a e essen ial o AI success
E en he bes AI ool needs human
knowledge. I ’s people who unde s and
cus ome s, p oduc s, and daily ope a ions.
Ou esea ch shows ha AI wo ks bes
when employees know how o use i and
can combine human expe ience wi h
digi al ools.
In in e iews and wo kshops, many wo ke s
said ha lea ning abou new echnologies
helps hem eel mo e confiden and alued.
They wan o unde s and how hese
sys ems wo k and how o use hem o make
hei jobs easie , no ha de .
As one pa icipan said:
“Talking o a cha bo can be us a ing –
bu when I know how i wo ks, I can help
cus ome s as e ins ead o figh ing he
sys em.”
6 | 249
WHAT SKILLS MATTER NOW
Re ail companies ac oss Eu ope a e
beginning o look o wo ke s who a e
com o able wi h echnology and open o
lea ning new ools. You don’ ha e o be an
IT specialis – bu some basic skills a e
becoming essen ial:
● Digi al li e acy: being com o able wi h
compu e s, able s, and mobile apps a
wo k.
● Unde s anding how da a is used:
knowing ha e e y ba code, pu chase,
and cus ome in e ac ion c ea es
in o ma ion ha helps he business un
be e .
● Communica ion skills: wo king wi h
bo h people and digi al sys ems, such as
POS ools, cha bo s, o CRM sys ems.
● P oblem-sol ing: no icing when
some hing goes w ong wi h an AI ool
and epo ing i in a cons uc i e way.
● Teamwo k and openness o lea ning:
sha ing expe iences, asking ques ions,
and helping each o he adap o change.
We call his combina ion o a i udes a
“digi al mindse ” – he abili y o s ay
cu ious, adap able, and posi i e abou
echnology, e en when i eels new o
unce ain.
WHAT ARE THE RISKS AND HOW TO
MANAGE THEM
Change can be di icul . Ou esea ch ound
ha many wo ke s wo y abou losing hei
jobs, being eplaced by machines, o being
judged by algo i hms. These ea s a e
unde s andable – and hey mus be
add essed openly.
The good news is ha mos companies
in oducing AI do no educe s a . Ins ead,
hey eo ganize oles so people can ocus
on highe - alue asks: cus ome ca e,
p oblem-sol ing, pe sonaliza ion, and
coo dina ion.
S ill, AI b ings challenges ha need
a en ion:
● Some ools a e no well designed and
make wo k ha de .
● No all employees ecei e p ope
aining.
● Communica ion be ween manage s and
eams abou digi al changes is
some imes weak.
Tha ’s why he INAiR p ojec calls o
inclusi e ans o ma ion – whe e wo ke s
a e in o med, ained, and in i ed o
pa icipa e in shaping how echnology is
used.
WHAT YOU CAN DO AS A RETAIL WORKER
● S ay cu ious. T y o unde s and wha
new ools do and how hey wo k. Ask
ques ions. Explo e how hey can make
you job easie .
● Build you digi al confidence. Lea n
basic compu e skills, how o use da a
ools, and how o in e ac wi h digi al
sys ems. Small s eps make a big
di e ence.
● Ask o aining. E e y company should
o e lea ning oppo uni ies. I you’ e no
su e how o use a new sys em, ask you
manage o suppo o sugges aking
pa in one o he INAiR lea ning
modules.
● Sha e you expe ience. Wo ke s know
bes whe e echnology helps and whe e
i ails. You eedback is essen ial o
imp o ing AI ools.
7 | 249
● P o ec wha ma e s. Be awa e o how
da a is collec ed and used. Make su e
ha au oma ion does no educe he
quali y o cus ome se ice o he digni y
o wo k.
HOW INAIR SUPPORTS WORKERS
The INAiR p ojec is de eloping ee,
p ac ical lea ning ma e ials o he e ail
sec o – a ailable online in se e al
languages. These Open Educa ional
Resou ces (OERs) a e designed o help
wo ke s like you:
● unde s and wha AI is and how i ’s used
in e ail,
● lea n he basics o da a and digi al ools,
● s eng hen p oblem-sol ing,
communica ion, and eamwo k,
● gain confidence o wo k wi h echnology
ins ead o agains i .
You can use hem indi idually o as pa o
company aining. The goal is o make su e
ha AI suppo s people, no he o he way
a ound.
THE MESSAGE FOR ALL RETAIL WORKERS
AI is pa o he u u e o e ail – bu ha
u u e is no fixed. I depends on how
people and o ganiza ions choose o use
echnology.
I wo ke s a e included, ained, and
espec ed, AI can make e ail jobs mo e
in e es ing, less epe i i e, and be e
paid. Bu ha equi es awa eness,
communica ion, and us be ween
employees, manage s, and o ganiza ions.
You a e no jus adap ing o he digi al
ans o ma ion – you a e pa o i .
You expe ience, c ea i i y, and empa hy
a e wha make e ail human. AI can help you
do you job be e – bu i canno eplace
wha only people can b ing: unde s anding,
connec ion, and ca e.
8 | 249
aining i benefi s a e clea .
● Manage s s uggle o connec AI o
business models, a he han ea ing i
as a “ ech add-on.”
● C oss-sec o coope a ion and suppo
s uc u es a e missing.
The wo kshops confi med ha he e ail
sec o ’s AI ansi ion equi es a new social
con ac – one ha balances inno a ion
wi h inclusion, and compe i i eness wi h
ai ness.
CHALLENGES THAT CALL FOR SYSTEMIC
RESPONSE
The ollowing sys emic ba ie s we e
epea edly iden ified ac oss esea ch
me hods:
● F agmen ed policy landscape: Di e en
suppo schemes exis bu a e no
aligned o easily accessible o SMEs.
● Unequal access o digi al
in as uc u e: small e aile s o en lack
a o dable da a ools and connec i i y.
● Skill sho ages and aining gaps: bo h
manage s and wo ke s lack s uc u ed
pa hways o build AI- ele an skills.
● Low us and anspa ency: e hical and
social conce ns emain un esol ed,
discou aging adop ion.
● Dominance o global digi al pla o ms:
Eu opean e aile s isk becoming
dependen on non-EU echnologies and
da a ecosys ems.
These challenges canno be sol ed by
indi idual companies alone. They equi e
collec i e s a egies ha combine
educa ion, policy alignmen , and indus ial
coope a ion.
RECOMMENDATIONS FOR SECTORAL
ORGANISATIONS AND POLICYMAKERS
1. C ea e sha ed aining in as uc u es.
Suppo he oll-ou o sec o -specific
lea ning p og ammes like he INAiR AI
Co e Cu iculum and OERs. Chambe s
and associa ions can help local e aile s
access hese ee esou ces and ailo
hem o hei needs.
2. P omo e skills isibili y.
Encou age companies o define and
communica e he AI- ela ed skills hey
equi e. This helps design be e
cu icula and align aining supply wi h
ma ke demand.
3. Build us ed knowledge ne wo ks.
Es ablish hubs o sha ing bes
p ac ices, case s udies, and pee
lea ning. Na ional chambe s and ade
associa ions can ac as in e media ies
be ween echnology p o ide s and small
e aile s.
4. Suppo social dialogue and wo ke
pa icipa ion.
Include unions and employee
ep esen a i es in discussions abou
au oma ion, e aining, and ask
edesign. Sha ed esponsibili y os e s
us and educes esis ance o change.
5. Ensu e ai compe i ion and da a
so e eign y.
Ad oca e o Eu opean solu ions and
in e ope abili y s anda ds ha p o ec
small e aile s om o e -dependence on
dominan global pla o ms.
6. In eg a e sus ainabili y and e hics.
Make su e AI adop ion aligns wi h g een
and social objec i es, p omo ing
esponsible inno a ion and decen wo k.
15 | 249

THE INAIR CONTRIBUTION: TOOLS FOR A
COORDINATED ECOSYSTEM
The INAiR p ojec o e s eady- o-use
ins umen s ha can se e as a ounda ion
o collabo a i e ini ia i es:
● AI Co e Cu iculum – a s uc u ed
lea ning amewo k o MSMEs in e ail,
ocusing on ans e sal, echnical, and
manage ial compe ences.
● Open Educa ional Resou ces (OERs) –
mul ilingual, accessible online modules
co-c ea ed wi h e aile s and expe s.
● Case s udies and bes p ac ices – 13
examples o AI implemen a ion in
Eu opean e ail MSMEs, illus a ing
di e se pa hs o inno a ion.
● Guidelines o cu iculum de elopmen
– adap able by sec o al o ganisa ions,
VET p o ide s, o chambe s o
comme ce.
These esou ces a e open-access and can
be in eg a ed in o na ional o egional
p og ammes o suppo capaci y building
and policy alignmen .
THE WAY FORWARD
AI in e ail is no jus a echnological shi –
i is a social and o ganisa ional
ans o ma ion ha will shape he u u e o
Eu opean comme ce.
Fo his ans o ma ion o be inclusi e,
compe i i e, and human-cen ed, i mus
be suppo ed by a s ong ecosys em:
● Companies willing o lea n and
expe imen ,
● Wo ke s empowe ed o adap and g ow,
● Ins i u ions and associa ions eady o
coo dina e and lead.
The INAiR p ojec p o ides he analy ical
basis and he p ac ical ools o such
coope a ion. Wha is now needed is
collec i e leade ship — whe e ade
o ganisa ions, chambe s, unions, and
educa ional pa ne s wo k hand in hand o
ensu e ha no e aile , wo ke , o
communi y is le behind in he age o AI.
16 | 249
EXECUTIVE SUMMARY FOR VOCATIONAL TEACHERS
AND STUDENTS IN THE RETAIL SECTOR
VET schools, aining cen es, and lea ne s p epa ing o e ail and cus ome se ice ca ee s
WHY AI MATTERS FOR VOCATIONAL
EDUCATION
The e ail and se ice indus ies a e
changing as e han e e be o e. Online
sales, sel -checkou sys ems, sma
shel es, da a-d i en ma ke ing, and
cus ome cha bo s a e no longe he u u e
– hey a e pa o e e yday business.
Behind mos o hese ools s ands a ificial
in elligence (AI): echnologies ha help
companies analyze da a, p edic demand,
and imp o e cus ome se ice. AI changes
how s o es a e managed, how eams a e
o ganized, and wha skills a e needed om
employees.
Fo oca ional schools and aining cen es,
his ans o ma ion c ea es a clea mission:
o p epa e lea ne s o jobs ha a e
inc easingly digi al, da a-d i en, and
collabo a i e.
The INAiR p ojec , unded by he Eu opean
Union, suppo s his goal by s udying wha
skills he e ail sec o eally needs and
c ea ing Open Educa ional Resou ces
(OERs) ha help eache s and s uden s
lea n abou AI in a p ac ical, accessible way.
WHAT THE RESEARCH SHOWS
The INAiR eam, led by esea che s om he
Uni e si y o Wa saw, s udied AI adop ion in
fi e coun ies: Cyp us, Ge many, I aly,
Poland, and Romania. The esul s highligh
a majo ansi ion ha is al eady unde way.
● AI adop ion is s ill low, especially in
small e ail businesses, bu i ’s g owing.
● Companies use AI mainly o imp o e
cus ome expe ience, manage s ock,
o ecas sales, and au oma e epe i i e
asks.
● The bigges ba ie is no echnology, bu
skills and confidence – many employees
and manage s s ill lack unde s anding o
wha AI can do.
● Digi al li e acy and ans e sal skills –
such as p oblem-sol ing,
communica ion, and adap abili y – a e
becoming as impo an as adi ional
echnical skills.
Fo oca ional educa ion, his means ha
aining mus combine p ac ical digi al
ools wi h c i ical and c ea i e hinking –
helping u u e wo ke s unde s and bo h
how AI wo ks and how o wo k wi h i
esponsibly.
WHAT SKILLS ARE NEEDED IN THE NEW
RETAIL LANDSCAPE
The epo iden ifies ou main g oups o
compe ences ha a e c ucial o he nex
gene a ion o e ail p o essionals:
17 | 249
● Digi al and AI li e acy – using digi al
de ices, apps, and da a ools
confiden ly; unde s anding how AI helps
in sales, logis ics, and cus ome suppo .
● Communica ion and eamwo k –
collabo a ing e ec i ely wi h colleagues
and echnology; sol ing p oblems ha
machines can’ .
● Adap abili y and lea ning mindse –
being eady o upda e skills as
echnology changes.
● E hical and social awa eness –
unde s anding how AI a ec s
cus ome s, jobs, and socie y, and using
i in a esponsible way.
Teache s can help lea ne s de elop hese
compe ences by in eg a ing AI- ela ed
examples in o e e yday lessons: om
analyzing online cus ome eedback o
expe imen ing wi h cha bo s o using basic
da a dashboa ds o ack sales ends.
HOW TEACHERS AND TRAINERS CAN USE
THE INAIR RESULTS
The INAiR epo and educa ional ma e ials
p o ide a eady- o-use ounda ion o
mode nizing oca ional cu icula in e ail
and ela ed fields.
He e’s how hey can suppo you:
● Unde s and indus y needs. The epo
explains wha skills employe s look o ,
which ools hey use, and whe e he
bigges knowledge gaps a e.
● B ing eal-li e examples o class. INAiR
collec ed 13 case s udies o companies
ha success ully implemen ed AI – om
amily- un s o es o online pla o ms.
These s o ies show ha AI is no only o
big ech companies bu also o small,
c ea i e businesses.
● Use open educa ional esou ces (OERs).
The p ojec is de eloping online,
mul ilingual lea ning modules ha help
eache s in oduce opics such as AI
basics, digi al ans o ma ion, cus ome
engagemen , and da a e hics.
● Encou age co-lea ning. S uden s and
eache s can lea n oge he – es ing
new ools, discussing eal p oblems, and
eflec ing on how echnology changes
wo k and skills.
This app oach helps schools s ay
connec ed o he eal needs o he labou
ma ke and ensu es ha lea ning emains
p ac ical, engaging, and u u e-o ien ed.
WHAT THIS MEANS FOR STUDENTS
Fo s uden s p epa ing o wo k in e ail,
hospi ali y, o cus ome se ice, he
message is clea : echnology will be pa o
you job, wha e e ole you choose.
Bu echnology will no eplace you. I needs
you c ea i i y, empa hy, and judgmen o
wo k well. To be eady, you can:
● lea n how digi al sys ems wo k and wha
da a hey use,
● build you confidence wi h ools like
sp eadshee s, cha bo s, o analy ics
dashboa ds,
● s ay cu ious abou new applica ions and
ask how hey help cus ome s and
cowo ke s,
● hink c i ically abou ai ness, p i acy,
and sus ainabili y in echnology use.
Employe s alue wo ke s who combine
p ac ical skills wi h cu iosi y and
esponsibili y. AI ools may change asks –
bu people who know how o lea n and
adap will always be in demand.
18 | 249
THE ROLE OF VOCATIONAL EDUCATION IN
A CHANGING EUROPE
The Eu opean Union is in es ing hea ily in
digi al upskilling and AI li e acy. Voca ional
educa ion plays a cen al ole in his
agenda, connec ing schools, employe s,
and local communi ies.
By in eg a ing AI- ela ed lea ning ou comes
in o hei p og ammes, VET ins i u ions
can:
● s eng hen hei ele ance o he labou
ma ke ,
● build pa ne ships wi h local businesses,
● help s uden s access quali y jobs in
g owing digi al sec o s,
● con ibu e o a ai , human-cen ed
digi al ansi ion.
The INAiR p ojec con ibu es o his e o
by c ea ing esou ces ha can be eely
used and adap ed by eache s, aine s, and
educa ional policymake s ac oss Eu ope.
THE TAKEAWAY: LEARN, ADAPT, AND
LEAD
AI is eshaping e ail, bu i is also c ea ing
new lea ning oppo uni ies.
Voca ional schools can be he b idge
be ween echnology and people – gi ing
young p o essionals he skills and
confidence o shape he u u e o wo k.
Teache s, aine s, and s uden s ha e a
sha ed mission: o make su e ha
echnology suppo s human c ea i i y, no
eplaces i . Lea ning abou AI oday is he
bes way o build secu e, meaning ul, and
inno a i e ca ee s omo ow.
19 | 249
INTRODUCTION
Aim o he Repo
The epo ep esen s he findings o esea ch conduc ed unde he auspices o he Wo k
Package 3 i led “Analysis o AI Skills Needs and Gaps in Re ail” o he “InAIR: Inc easing he
Up ake o AI in Re ail” p ojec , Coo dina ion and Suppo Ac ion unded by he Eu opean Union'’s
Ho izon Eu ope Resea ch and Inno a ion p og amme unde G an Ag eemen No. 101133847.
The aim o his epo is o ulfill he fi s specific objec i e o he p ojec , which is o iden i y
he AI skill needs and gaps o MSMEs in he e ail sec o in Cyp us, Ge many, I aly, Poland and
Romania.
This esea ch aims o guide e ail sec o companies and hei employees h ough he digi al
e olu ion. Ou mo o is well eflec ed in he wo ds o one o he e ail sec o s akeholde s who
sha ed hei knowledge wi h us.
“
I seems we a e always
a aid o he unknown,
akin o hesi a ing a he
edge o a da k o es .
Ye , e e y b eak h ough
and e olu ion in ol es
s epping in o ha
da kness, ge ing o
know i , and hen mo ing
beyond.
In _Exp_Poland_9
20 | 249

Main esul s
AI adop ion in e ail is
unde - esea ched
Resea ch on he impac o AI on he e ail
sec o 's ope a ions is s ill in i s ea ly s ages.
The exis ing li e a u e lacks comp ehensi e
s udies on he ole o AI skills in small and
medium-sized en e p ises wi hin he e ail
sec o , especially conce ning sus ainable
de elopmen and g een skills (see Sec ion 1
and Sec ion 4).
Re ail companies exhibi a
low le el o AI adop ion
The e ail sec o aces dis inc challenges
in i s digi al ans o ma ion jou ney, no ably
wi h AI in eg a ion, due o sec o al
agmen a ion (97% o companies a e
mic o, small, and medium-sized
en e p ises), low le els o digi aliza ion
(especially in Poland and Romania), a slow
a e o AI adop ion (pa icula ly in Poland),
and insu icien digi al skills among
employees (see Sec ion 2 and Sec ion 7).
Employe s do no seek AI
skills (ye )
The limi ed p og ess in AI in eg a ion wi hin
e ail companies is e iden om employe s'
minimal in e es in hi ing AI-skilled wo ke s
(see Sec ion 5).
Re ail companies adop AI
echnologies in di e se ways
Companies demons a e di e se
app oaches o using and in eg a ing AI
echnologies (see Sec ion 3 and Sec ion 6).
Consequen ly, he AI Co e Cu iculum and
open educa ional esou ces should accoun
o hese a ied models. Simila ly,
educa ional s a egies should ca e o
specific subse s o ans e sal skills
equi ed by e ail company manage s, such
as change managemen compe encies,
leade ship skills, he abili y o o mula e
g ow h s a egies, and comp ehensi e
knowledge o AI applica ions and AI p ojec
managemen (see Sec ion 8).
Digi al mindse is a new
ans e sal skill
Digi al ans o ma ion necessi a es a new
se o ans e sal skills and compe encies,
e e ed o as a "digi al mindse "
cha ac e ized by openness o inno a ion
and o ganiza ional change, collabo a ion
wi h new echnologies and AI, and
undamen al echnical skills (see Sec ion 8).
21 | 249
O e iew o he Repo
Sec ion 1 o his epo examines he po en ial
o AI echnology in he e ail sec o . We
conduc ed a sys ema ic and quali a i e
li e a u e e iew, complemen ed by insigh s
om expe s and s akeholde s ga he ed
h ough co-c ea ion wo kshops and
ques ionnai es. The academic discou se, led
by STEM esea che s, p ima ily ocuses on
echnical ma e s such as decision suppo
sys ems, na u al language p ocessing, and he
In e ne o Things. Howe e , he e is a lack o
in-dep h s udies on AI skills in MSMEs and
sus ainabili y. This examina ion o AI in e ail
conside s bo h echnical and socio-cul u al
pe spec i es, ocusing on compu a ional
ad ancemen s and e hical conside a ions. AI
has he po en ial o ans o m he e ail sec o
by enhancing decision-making, au oma ing
p ocesses, pe sonalizing cus ome
in e ac ions, and op imizing wo k o ce
managemen . Despi e hese ad an ages, AI
also aises e hical and social conce ns. One
conce n is he po en ial o AI o be pe cei ed
as ha ing men al capaci ies, which could
complica e he oles o humans and
machines. Addi ionally, AI-d i en au oma ion
poses a isk o job displacemen . The e o e, i
is c ucial o adop a p oac i e app oach o
managing he socio-economic impac s o
digi aliza ion.
Sec ion 2 analyzes he s a e o AI adop ion in
he e ail sec o , ocusing on selec ed
Eu opean Union coun ies: Cyp us, Ge many,
I aly, Poland and Romania. O e he pas
decade, he wholesale and e ail ade sec o
has no ably expanded, wi h Romania and
Poland showing he highes u no e , while
Ge many and I aly ha e lagged behind he EU
a e age, indica ing dispa i ies in g ow h a es
ac oss Eu ope. The sec o is domina ed by
mic o, small and medium-sized en e p ises
(MSMEs), especially in Romania, which hinde s
compe i i eness and ope a ional e iciency,
hus slowing digi aliza ion. The e ail sec o
lags behind indus ies such as in o ma ion
and communica ion echnology (ICT) and
p o essional se ices in e ms o
digi aliza ion. Ge man and Cyp io en e p ises
pe o m abo e he EU a e age, while Polish
fi ms exhibi low le els o digi aliza ion. The
a ia ion in e-comme ce adop ion indica es a
dispa i y in digi al in eg a ion le els. AI
adop ion in he e ail sec o is he lowes
among he analyzed sec o s, wi h Ge many
exhibi ing he lowes a e a 10%, and Poland
ha ing he highes pe cen age o fi ms no
using AI. AI applica ions a e commonly used in
obo ic p ocess au oma ion, speech
ecogni ion, and ex mining. Cloud
echnology adop ion is mos p e alen in
Cyp us and I aly, while Romania has he lowes
a es. Simila ly, he implemen a ion o he
In e ne o Things (IoT) a ies conside ably
ac oss he egion. The adop ion o AI is
hinde ed by a lack o specialized knowledge,
high cos s, and legal unce ain ies. Low
digi al skills, pa icula ly in Poland and
Romania, u he impede he adop ion o
echnology in he e ail sec o and MSMEs.
In Sec ion 3, we p esen a se ies o case
s udies and bes p ac ices illus a ing he
eal-wo ld applica ions o AI in he e ail
sec o . Ou sou ced AI ools, equen ly
ma ke ed as equi ing no echnical expe ise
o main enance, in ac necessi a e a
undamen al comp ehension o a company's
ope a ional da a and he AI ool's unc ionali y
and ad an ages. The implemen a ion o hese
ools necessi a es he exe cise o c i ical
hinking skills o e alua e hei pe o mance.
In mos cases, AI is employed o enhance
cus ome suppo , unde sco ing he necessi y
22 | 249
o o ganiza ions o possess a ounda ional
unde s anding o AI and i s applica ions.
Sec ion 4 examines he defini ions and
classifica ions o AI skills. Cu en ly, he e is a
lack o comp ehensi e s udies on AI skills in
MSMEs, pa icula ly in he e ail sec o , and on
sus ainabili y and g een skills. Mos AI
esea ch in he e ail sec o is conduc ed by
STEM p o essionals, ocusing on echnical
aspec s such as decision suppo sys ems,
na u al language p ocessing, and he In e ne
o Things. Recen ini ia i es emphasize
li e acy, social, and echnical compe encies.
AI applica ions in he e ail sec o include
business decision suppo , au oma ion o
in en o y and supply chain p ocesses,
enhancemen o cus ome ela ionships,
moni o ing consume beha io , and
au oma ing ec ui men p ocedu es. Po en ial
challenges associa ed wi h AI in e ail include
job losses, a lack o sui able skills, us
issues, echnical di icul ies, and he need o
sus ainable p ac ices. The absence o a
sys ema ic defini ion o he ela ionship
be ween AI skills and digi al ans o ma ion in
he e ail sec o highligh s he need o
empi ical esea ch. The OECD Employmen
Ou look epo p oposes a amewo k o AI
skills equi ed o he de elopmen ,
main enance, and in e ac ion o AI sys ems.
Add essing hese gaps could enhance ou
unde s anding o he in e connec ions
be ween AI skills, digi al ans o ma ion, and
sus ainabili y in he e ail sec o . A esea ch
agenda in eg a ing quali a i e and
quan i a i e me hods and mul iple
s akeholde pe spec i es is essen ial.
Sec ion 5 p esen s an analysis o he u u e AI
skills ha employe s will equi e. The key
findings include a compa a i e analysis o job
ad e isemen s ac oss selec ed coun ies,
which e ealed se e al commonali ies. These
include a p onounced emphasis on
p oficiency in o ice so wa e such as
Mic oso O ice and en e p ise esou ce
planning (ERP) sys ems such as SAP, along
wi h a obus demand o cus ome
ela ionship managemen (CRM) skills and
expe ise in digi al ma ke ing (e.g., Google
Ads) and e-comme ce pla o ms.
Fu he mo e, he epo indica es ha basic
compu e skills and o e all digi al li e acy a e
conside ed essen ial. Ne e heless, i is
e iden ha he e a e no able di e ences
be ween coun ies. Cyp us places a p emium
on agile me hodologies and design ools such
as Figma and Adobe Design. Ge many, on he
o he hand, places a s ong emphasis on cloud
echnologies (AWS, Docke ), ad anced IT
in as uc u e, and c ea i e skills (Adobe
Pho oshop, Illus a o ). I aly, meanwhile,
alues da a business in elligence, social
media managemen , and ha dwa e ins alla ion
and suppo . Poland, finally, places a
significan ocus on ad anced Excel skills and
specific SAP modules. Romania, on he o he
hand, places a s ong emphasis on eme ging
echnologies such as VR glasses and
elec onic p ocessing so wa e. Mo eo e ,
I aly and Romania place a s ong emphasis on
cus ome in e ac ion and e ail on -end
compe encies, while Cyp us and Ge many
end o p io i ize echnical and backend skills.
Sec ion 6 examines he AI skills equi emen s
and deficiencies om he pe spec i e o
s akeholde s in he e ail sec o . This sec ion
examines he challenges cu en ly acing he
indus y and conside s how a ificial
in elligence can o e po en ial solu ions. In a
se ies o co-c ea ion wo kshops held ac oss
selec ed EU coun ies, pa icipan s iden ified
a se o essen ial skills and compe encies ha
mic o, small and medium-sized en e p ises
(MSMEs) should possess o le e age AI
e ec i ely. Fu he mo e, hey iden ified skills
23 | 249
ha a e essen ial o manage s and
employees. This sec ion p esen s an o e iew
o he skill se s ha we e collec i ely
iden ified du ing he wo kshops. The
subsequen sec ion delinea es he equisi e
skills and cha ac e is ics ha MSMEs mus
possess o success ully implemen a ificial
in elligence (AI). This is ollowed by a
discussion on he c i ical AI skills ha a e
necessa y o bo h business leade s and
employees.
Sec ion 7 examines he obs acles impeding
he implemen a ion o AI in he e ail sec o ,
wi h a pa icula ocus on he challenges
aced by MSMEs. P e ious s udies ha e
iden ified nume ous ba ie s o he adop ion
o AI in he e ail sec o . These include
conce ns o e job displacemen , a sca ci y o
AI expe ise, and issues su ounding us
bo h in e nally and wi h cus ome s.
Fu he mo e, echnical obs acles, conce ns
ega ding sus ainabili y, and he absence o
indus y s anda ds p esen significan
challenges. The co-c ea ion wo kshop
iden ified se e al key obs acles aced by
MSMEs in he e ail sec o . These include
main aining e icien supply chains and
managing in en o y, na iga ing digi al
ans o ma ion, adap ing o AI-d i en
echnological shi s, and e ec i ely
add essing communica ion challenges in
cus ome ela ionships spanning ma ke ing
and se ice. Pa icipan s emphasized he
compe i i e p essu es exe ed by digi al
gian s om he Uni ed S a es and China,
which ha e a p o ound impac on he sec o 's
landscape. Fu he mo e, he wo kshop
highligh ed he di icul y ha MSMEs
encoun e in e aining and nu u ing a skilled
wo k o ce capable o handling AI-d i en
ad ancemen s.
The objec i e o Sec ion 8 is o p esen
ecommenda ions and guidelines o he
de elopmen o an AI co e cu iculum ha is
specifically ailo ed o MSMEs ope a ing
wi hin he e ail sec o . The discussions held
du ing he co-c ea ion wo kshops, which
we e conduc ed in a ious EU coun ies,
ocused on he planned aining ini ia i es
ha we e designed o enhance he AI skills
wi hin he e ail sec o . This sec ion is based
on he insigh s sha ed by pa icipan s and
p oposes a s uc u ed app oach ha
p io i izes p ac ical ad ice. The epo
ou lines he specific needs iden ified du ing
he wo kshops in o de o ensu e ha he
cu iculum e ec i ely add esses he
challenges and equi emen s ha a e unique
o MSMEs in he e ail sec o .
24 | 249
helped o iden i y he main opics add essed wi hin he field, cha ac e ize he leading
academic jou nals, and ou line he p ima y esea ch a eas discussing AI applica ions in e ail.
1.1.1. Da a Collec ion
In he da a collec ion phase, we used Scopus – a leading ci a ion da abase enowned o i s
comp ehensi e eposi o y o high-quali y, pee - e iewed pape s (Zhu and Liu, 2020). In he
e ie al p ocess, we used he Scopus API o sys ema ically ex ac he ele an publica ions
and hei me ada a.
The keywo ds combina ions we used we e:
TITLE-ABS-KEY ( X AND Y ) AND PUBYEAR > 2017 AND PUBYEAR < 2025
Whe e: X=['sales', ' e ail', 'e-comme ce'], Y=['AI', 'a ificial in elligence']. These keywo ds we e
sc u inized wi hin i les, abs ac s, au ho keywo ds, and ‘ opics’ as delinea ed by he pla o m.
To ensu e he eliabili y o ou analysis, we efined ou ocus exclusi ely o pee - e iewed
a icles, specifically limi ing ou scope o he ca ego ies o a icles and e iews.
Fo ep oducibili y pu poses we p esen he exac sea ch que y:
(TITLE-ABS-KEY ( "AI" ) OR TITLE-ABS-KEY ( "a ificial in elligence" ) ) AND ( TITLE-ABS-KEY (
"sales" ) OR TITLE-ABS-KEY ( " e ail" ) OR TITLE-ABS-KEY ( "e-comme ce" ) ) AND ( DOCTYPE (
a ) OR DOCTYPE ( e ) ) AND PUBYEAR > 2017 AND PUBYEAR < 2025
This sea ch yielded a o al o 1699 publica ions (a e duplica es emo al). In he analysis we
used ull ex s o 645 a icles o which we had pe missions o e ie e pd s and me ada a and
abs ac s o he ull sample.
1.1.2. Resul s
The final da ase includes a o al o 1,604 esea ch a icles and 95 e iews (schola ly pape s
summa izing he cu en s a e o esea ch). These a icles co e esea ch om 105 di e en
coun ies, as indica ed by au ho s a ilia ions, demons a ing a wide geog aphical dis ibu ion.
The o al numbe o au ho s is 5532 and hey ep esen 273 dis inc ins i u ions, and hei wo k
is published in a o al o 809 di e en jou nals. A e age collabo a ion index (numbe o au ho s
pe a icle) is 3.42 (single au ho pape s cons i u e only 13% o o al) meaning ha esea ch in
his a ea is o en done in collabo a ions. A e age ci a ion pe documen is 19 which is a good
sco e conside ing ha we analyze ecen esea ch.
In Table 2 we p esen he mos common keywo ds p o ided by au ho s o classi y hei
esea ch. We used op 50 keywo ds om which we emo ed he i ial ones (e.g. AI,
e-comme ce).
31 | 249

Table 2 Top au ho keywo ds
deep lea ning
co id-19
big da a
in e ne o hings
na u al language p ocessing
indus y 4.0
cha bo
social media
da a mining
inno a ion
blockchain
digi al ma ke ing
decision suppo sys ems
supply chain managemen
sen imen analysis
ecommende sys ems
decision suppo sys em
pu chase in en ion
digi aliza ion
o ecas ing
cus ome expe ience
au oma ion
us
p edic ion
io
pe sonaliza ion
explainable ai
cha gp
ma ke ing
se ice obo s
cus ome se ice
in e ne o hings (io )
classifica ion
da a analy ics
neu al ne wo ks
cus ome sa is ac ion
sus ainabili y
cus ome ela ionship managemen
supply chain
compu e ision
digi al ans o ma ion
Sou ce: own elabo a ion
In Table 3 we p esen he subjec a eas (b oad disciplines) o ca ego ize he esea ch on AI and
e ail. Table 4 shows main jou nals in which au ho s publish his esea ch.
Table 3 Top subjec a eas
Subjec a ea
Numbe o publica ions
Compu e Science
960
Enginee ing
569
Business, Managemen and Accoun ing
504
Social Sciences
312
Decision Sciences
192
32 | 249
Ma hema ics
185
Economics, Econome ics and Finance
139
Ma e ials Science
111
Psychology
89
A s and Humani ies
83
En i onmen al Science
81
Sou ce: own elabo a ion
Table 4 Top jou nals
Jou nal Name
Numbe o Publica ions
Sus ainabili y
34
IEEE Access
31
Technological Fo ecas ing and Social Change
22
Compu e s and Indus ial Enginee ing
18
Expe Sys ems wi h Applica ions
17
Applied Sciences
17
In e na ional Jou nal o In o ma ion Managemen
16
Mobile In o ma ion Sys ems
15
Elec onics (Swi ze land)
15
Decision Suppo Sys ems
14
Sou ce: own elabo a ion
The selec ed scien ific a icles a e domina ed by hose om he fields lis ed in Table 3, i.e.
compu e science and enginee ing, ollowed by hose om managemen and social sciences.
Wi hin he o al pool o a icles, hose wi h cha ac e is ics simila o hose o he STEM fields
domina e, wi h ewe a icles om he social sciences and humani ies. I is no ewo hy ha he
a icles lis ed include hose w i en in he field o en i onmen al sciences. In e ms o jou nals,
he mos impo an jou nals lis ed in Table 4 a e Sus ainabili y, IEEE Access and Technological
Fo ecas ing and Social Change. As in he fields, he jou nals a e domina ed by jou nals wi h a
echnical ocus, while jou nals in he social sciences and humani ies play a mino ole o , as in
he case o Sus ainabili y, a e o low quali y.
We also de ailed he mos common wo ds and ph ases in he a icle abs ac s as pa o ou
sys ema ic analysis o he exis ing li e a u e on AI in e ail, esul ing in Figu e 2 and Table 5.
Figu e 2 Wo dcloud o mos common e ms appea ing in a icles' abs ac s
33 | 249
Sou ce: own elabo a ion
Table 5 Top 3-g ams appea ing in a icles’ abs ac s
3-g am
Numbe o occu ences
decision suppo sys ems
114
na u al language p ocessing
58
in e ne o hings
56
an colony op imiza ion
53
supply chain managemen
45
a ificial in elligence decision
45
a ificial in elligence echnologies
44
language p ocessing sys ems
36
classifica ion o in o ma ion
35
decision making decision
31
lea ning algo i hms lea ning
31
algo i hms lea ning sys ems
30
in elligence decision making
28
sales a ificial in elligence
27
making decision suppo
26
suppo ec o machines
26
algo i hms machine lea ning
25
a icle a ificial in elligence
25
a ificial in elligence da a
23
deep neu al ne wo ks
23
34 | 249
lea ning algo i hms machine
23
quali y o se ice
22
language p ocessing na u al
22
op imiza ion a ificial in elligence
21
a ificial in elligence comme ce
21
a ificial in elligence beha io al
20
cus ome ela ionship managemen
19
colony op imiza ion a ificial
19
machine lea ning algo i hms
19
in elligence beha io al esea ch
19
pa icle swa m op imiza ion
19
a ificial in elligence consump ion
18
in elligence consump ion beha io
18
a ificial in elligence big
18
in elligence big da a
18
human compu e in e ac ion
17
sales sen imen analysis
17
long sho e m
17
sho e m memo y
17
con olu ional neu al ne wo k
17
public ela ions sales
16
5g mobile communica ion
16
a ificial in elligence cus ome
16
deep lea ning lea ning
16
swa m op imiza ion pso
16
in o ma ion and communica ion
16
a ificial in elligence elec onic
16
design me hodology app oach
15
algo i hm a ificial in elligence
15
con olu ional neu al ne wo ks
15
Sou ce: own elabo a ion
Among he op 3 g ams lis ed in Table 5, he op 10 issues wo h no ing a e decision suppo
sys ems, na u al language p ocessing, he In e ne o Things, an colony op imiza ion, supply
chain managemen , a ificial in elligence decision, a ificial in elligence echnologies, language
p ocessing sys ems, classifica ion o in o ma ion, and decision-making decision. The analysis
shows ha he li e a u e o da e on he applica ion o AI in e ail has mainly ocused on he
echnical aspec s o applying his echnology, i s ma e ial conside a ions ela ed o
in as uc u e and cloud compu ing, as well as he managemen aspec s o using his ype o
35 | 249
echnology. Figu e 2 p esen s he mos common single wo ds and Table 5 combina ions o
h ee wo ds (so called 3-g ams) used in a icles’ abs ac s. The main in e es has been limi ed o
picking up he h ead o models applied o cus ome a eas, as he cloud wi h he mos
equen ly used wo ds migh sugges .
Mo eo e , we employed La en Di ichle Alloca ion (LDA) using BERTopic o iden i y opics
wi hin he abs ac s o a icles ocused on AI in e ail. LDA is a gene a i e s a is ical model ha
allows se s o obse a ions o be explained by unobse ed g oups, e ec i ely unco e ing he
hidden hema ic s uc u e in a collec ion o documen s (Jeloda e al., 2019). BERTopic, on he
o he hand, le e ages BERT embeddings o documen ep esen a ion and clus e ing,
enhancing he in e p e abili y and cohe ence o he opics disco e ed (G oo endo s , 2022).
By applying LDA wi h BERTopic, we we e able o sys ema ically o ganize he esea ch a icles
in o dis inc opics, acili a ing easie na iga ion and analysis o he li e a u e. The esul ing
model has been sa ed in ou p ojec eposi o y on Zenodo. This file allows esea che s wi h
basic Py hon skills o an o ganized e iew o esea ch and enables b owsing a icles by specific
opics. In Table 6 we p esen he opics along wi h he mos cha ac e is ic keywo ds and
numbe o a icles o which he gi en opic is a main one. Names o opics we e c ea ed using
Cha GPT4o based on ull lis s o cha ac e is ic keywo ds.
36 | 249

Table 6 Topic modeling esul s
Topic Name
Keywo ds
A icle
Coun
AI Cha bo s and Consume
In e ac ion
[ai, cha bo s, cha bo , consume s]
215
P oduc Re iews and Sen imen
Analysis
[ e iews, sen imen , p oduc , analysis]
62
IoT and Cloud Compu ing
[io , cloud, da a, sma ]
60
Recommenda ion Sys ems
[ ecommenda ion, ecommende ,
ecommenda ions]
55
Digi al T ade and Economic
T ans o ma ion
[digi al, ade, coun ies, ans o ma ion]
49
Cus ome Da a Mining and Beha io
[da a, cus ome , mining, beha io ]
43
Fashion Design and Image
P ocessing
[ ashion, design, image, clo hing]
43
Supply Chain Decision Making
[decision, supply, chain, in en o y]
39
Demand Fo ecas ing and Sales
Models
[ o ecas ing, demand, models, sales]
37
Ag icul u al P ac ices and Poul y
Fa ming
[poul y, a me s, ag icul u al, u al]
35
Se ice Robo s and Cus ome
Accep ance
[ obo s, se ice, obo , accep ance]
26
Cus ome Loyal y and Me a e se
[cus ome , me a e se, loyal y, ools]
25
Omnichannel Re ail and AI
[ e ail, ai, e aile s, omnichannel]
23
Indus ial Supply Chain Managemen
[chain, supply, manu ac u ing, indus ial]
22
Cus ome Chu n P edic ion
[chu n, cus ome , models, p edic ion]
21
Sma Re ail Sec o
[ e ail, e ailing, sma , sec o ]
19
Logis ics and E-comme ce
Dis ibu ion
[logis ics, dis ibu ion, comme ce, c oss]
19
Business Ma ke ing and
E-comme ce Resea ch
[comme ce, ma ke ing, business,
esea ch]
18
Financial C edi Risk and Sco ing
[c edi , isk, financial, sco ing]
18
AI in Sales and Selling Techniques
[sales, salespeople, ai, selling]
16
F aud De ec ion in T ansac ions
[ aud, de ec ion, ansac ions,
audulen ]
15
Use Models in Mobile Comme ce
[use , model, ained, mobile]
10
Sou ce: own elabo a ion
37 | 249
Among he hemes de ailed in he sys ema ic li e a u e e iew in Table 6, he issue o
consume in e ac ion wi h cha bo s based on a ificial in elligence echnology eme ges as a
key a ea o ocus. This includes how consume s engage wi h AI-powe ed cha bo s and he
implica ions o cus ome se ice and sa is ac ion. In addi ion, consume ela ionships ex end
o se e al o he c i ical aspec s.
Fi s ly, p oduc e iews and ma ke ing sen imen analysis a e highligh ed as key issues. This
in ol es he use o AI o analyze consume eedback and sen imen exp essed in p oduc
e iews, helping e aile s o unde s and consume opinions and imp o e hei p oduc s and
se ices.
Secondly, ecommenda ion sys ems a e iden ified as ano he impo an a ea. AI-powe ed
ecommenda ion sys ems enhance he consume shopping expe ience by sugges ing p oduc s
based on pas pu chase beha io and p e e ences. Consume da a mining and he analysis o
consume shopping beha io also appea as cen al hemes in he li e a u e. These hemes
explo e how AI echnologies can p ocess as amoun s o consume da a o iden i y pa e ns
and ends, which can hen be used o ailo ma ke ing s a egies and op imize in en o y
managemen .
S uc u al issues ela ed o da a in as uc u e and cloud compu ing also ecei e significan
a en ion. These hemes del e in o he echnological backbone equi ed o suppo AI
applica ions in e ail, highligh ing he impo ance o a obus da a in as uc u e and he ole o
cloud compu ing.
Finally, supply chain managemen is ano he key heme in he li e a u e. The ole o AI in
op imizing supply chain ope a ions, om in en o y managemen o logis ics and dis ibu ion, is
explo ed, showing how AI can imp o e e iciency and educe cos s in he e ail sec o .
1.2. AI uses in e ail based on quali a i e li e a u e e iew
In his sec ion, we discuss he challenges o defining a ificial in elligence in e ail as
p esen ed in a ious li e a u e on he subjec . We also highligh how AI is defined in e ms o
i s unc ions and applica ions in he e ail indus y.
1.2.1. Defini ions o AI
A ificial in elligence is b oadly desc ibed as he abili y o a digi al compu e o
compu e -con olled obo o pe o m asks ypically associa ed wi h in elligen beings. The
e m o en e e s o e o s o c ea e sys ems ha possess human-like in ellec ual abili ies,
such as easoning, unde s anding meaning, gene alizing, and lea ning om pas expe ience
38 | 249
(Go don, 2023: 16). Howe e , as Fehe and Ka ona (2021) poin ou , a ificial in elligence is
unde s ood in wo gene al ways: in a echnical way and in a socio-cul u al way.
The dominan cu en in defining AI is o a echnical na u e. Technical unde s anding o AI
e e s o a ca ego y o compu a ional echnologies ha d i e AI ad ancemen s, emphasizing
he STEM fields – science, echnology, enginee ing, and ma hema ics – ha unde pin
inno a ions like obo ic au oma ion and machine lea ning (Fehe and Ka ona, 2021: 1). This
app oach ocuses on using AI echnology ac oss a ious indus ies, wi h he de elopmen and
applica ion o AI se ing as he ul ima e goal o achie e a compe i i e ad an age (Fehe and
Ka ona, 2021: 1). Ne e heless, as he au ho s poin ou , a second s and is also eme ging in
defining AI, ocusing on he socio-cul u al unde s anding o AI. The second p oposed e m
ocuses on he use o echnology as a means a he han an end in i sel . This ca ego y
emphasizes he inco po a ion o AI echnology in o social and cul u al con ex s, conside ing
ac o s such as human adap a ion and e hical issues. The socio-cul u al AI pe spec i e is
suppo ed by inc easing a en ion o public and co po a e policy, cul u al no ms, and AI e hics,
highligh ing he b oade implica ions o socio-cul u al echnology (Fehe and Ka ona, 2021: 2).
To accoun o he social uses o AI in di e en con ex s, including e ailing con ex s, we should
ocus on he socio-cul u al s and o AI defini ion. When i comes o defining AI by i s
social-alike cha ac e is ics, Caluo i (2023: 4-6) has o mula ed a se o defini ional c i e ia:
● Lea ning Abili y: lea ning abili y is a key c i e ion o AI, pa icula ly because machine
lea ning is equen ly used as a synonym o AI;
● Human Likeness: he esemblance o human beings is a majo ac o ha makes AI
in iguing, bo h in scien ific and popula con ex s. This c i e ion is he mos equen ly
used in AI defini ions, al hough i s applica ion a ies;
● S a e o Mind: he e olu ion o AI can also be seen in he “s a e o mind” dimension
unde s ood as gi ing echnology an in ellec ual s a e;
● Sol ing Ha d P oblems: he complexi y o he p oblem which AI is able o sol e is he
leas con en ious c i e ion, as mos defini ions do no conside i essen ial;
● Success ul Solu ions o P oblems: he c i e ion o success ulness shows a dis ibu ion
like he "s a e o mind" dimension, indica ing i is also con en ious.
The abo e se o c i e ia o defining a ificial in elligence has implica ions o how AI
echnologies a e iewed in di e en social, cul u al and economic con ex s. The field o e ail
and s udies o he applica ion o AI in e ail eflec all o he abo e c i e ia. Conside he
ollowing defini ion o AI in e ail p o ided by Gi oux e al. (2022: 1028):
One echnology ha companies (..) ha e s a ed o emb ace is AI, which e e s o
“p og ams, algo i hms, sys ems and machines ha demons a e in elligence”
(Shanka , 2018, p. i) and is “mani es ed by machines ha exhibi aspec s o human
39 | 249
in elligence” (Huang & Rus , 2018, p. 155). Wi h i s abili y o accu a ely pe o m asks
and goals based on ex e nal inpu s, AI is e olu ionizing how companies and
o ganiza ions c ea e con en , make ecommenda ions, and in e ac wi hin he s o e
(de Ruy e e al., 2018; Haenlein & Kaplan, 2019; Webe & Schü e, 2019). Robo s a e
also p omising a enues o on line se ices. Fo he momen , obo s a e mainly
implemen ed in he manu ac u ing and deli e y o p oduc s and se ices (I ano ,
2020), bu we can expec hei g owing p esence in on line se ices.
In he defini ion abo e, AI is defined as compu e p og ams in he b oades sense ha exhibi
in elligence and in his espec esemble human in elligence. AI echnologies a e also defined
in e ms o hei abili y o sol e p oblems and achie e goals, i.e. o success ully p esen
solu ions o specific p oblems. In he quo e abo e, he au ho s also ou line he po en ial
applica ions o AI in e ail, wi h a ocus on con en p oduc ion and in-s o e ecommenda ions.
Mo eo e , a ious ypes o cha bo s and i ual assis an s a e being used in he e ail sec o ,
capi alizing on hei abili y o mimic human in e ac ions. Designe s can s a egically
inco po a e elemen s such as small alk, g ee ings, and con e sa ional ansi ions o build use
us in he in e ace and encou age specific beha io s, including sel -disclosu e and
pe suasion (Schanke e al., 2021: 6). Howe e , he use o cha bo s in cus ome in e ac ions
b ings wi h i addi ional dilemmas – which a e discussed mo e in he sec ion on co-c ea ion
wo kshops. As desc ibed by Gi oux e al. (2022: 1030), he esea ch highligh s ha he
a ibu ion o men al capaci ies o echnologies significan ly influences mo al judgemen s and
no ions o esponsibili y, bu hey cau ion agains di ec compa isons wi h human- o-human
in e ac ions, ul ima ely emphasizing he peculia i ies o he human mind and he challenges o
applying human no ms and mo al esponses o machines.
I is impo an o no e ha we a e no dealing wi h a single a ificial in elligence, bu many
di e en ones ha mani es hemsel es in specific cus ome in e ac ion ools. Pan ano and
Sca pi (2022: 586) p esen ed he ollowing ypology o o ms o a ificial in elligence used in
cus ome in e ac ion:
● Mechanical o ope a ional in elligence: he abili y o lea n and pe o m basic and
epe i i e asks.
● Thinking: he abili y o pe o m analy ical and in ui i e asks ( easoning-based
in elligence).
● Emo ional o a ec i e: he abili y o ecognize human emo ions and adap beha io
acco dingly.
● Sel -o ganizing coope a ion: he abili y o coo dina e wi h o he AI o c ea e a
sel -managed, au onomous, collabo a i e ne wo k (dis ibu ed in elligence).
● Social cogni ion: he abili y o p ocess, s o e, and apply in o ma ion abou o he s and
beha e acco dingly.
40 | 249
B. Cus ome engagemen
When in e ac ing wi h a company's sales o communica ion channels, cus ome s may
be willing o benefi om op imized expe ience, bu a he same ime dis anced om
he idea o being guided by ully au oma ed se ices. Augmen ed in elligence o e s help
h ough p oblem-sol ing and pe sonalized ecommenda ions wi hou dep i ing a clien
o he sense o con ol o e decision-making. I can con ex ualize a pe son's in en ions
and balance be ween he su iciency o cha bo s and he need o human suppo .
Whene e sui able and desi ed, augmen ed in elligence pe o ms complex analy ical as
well as he unengaging epe i i e asks, eeing up human labo on bo h sides o he
shopping expe ience.
C. In en o y managemen
Augmen ed in elligence suppo can help o ganize an e icien supply chain. I enables
he employees o easily access he de ailed in o ma ion on p edic ed demand and how
he in en o y main enance, as well as o he supply- ela ed p ocesses, can be adjus ed
o mee i in he op imal way. I is able o o e see he complex da a and p o ide
con ex -specific epo s wi hou independen ly in e e ing wi h he company’s
ope a ions.
D. Ope a ions op imisa ion
Wi h augmen ed in elligence sys ems aid, a company's p oduc i i y is no as limi ed by
i s labo o ce capabili ies. I may be easie o he en e p ise o scale up i s ope a ions,
when some o he ac i i ies a e handled h ough in elligen solu ions. Lowe - anked
employees can become esponsible o pe o ming mo e ad anced asks wi hou he
need o ex ensi e supe ision, which in u n may con ibu e o inc eased us inside he
ins i u ion.
1.2.3.4. Insigh engines
Insigh engines “apply ele ancy me hods o desc ibe, disco e , o ganize and analyze da a. This
allows exis ing o syn hesized in o ma ion o be deli e ed p oac i ely o in e ac i ely, and in he
con ex o digi al wo ke s, cus ome s o cons i uen s a imely business momen s. P oduc s in
his ma ke use connec o s o c awl and index con en om mul iple sou ces. They index he
ull ange o en e p ise con en , om uns uc u ed con en such as wo d p ocesso and ideo
files h ough o s uc u ed con en , such as sp eadshee files and da abase eco ds.” (Ga ne ,
2024c). Insigh engines, backed by algo i hms and na u al language p ocessing, a e able o
47 | 249

p ocess mo e nuanced and di e se da a han adi ional sea ch engines, iden i ying ele an
in o ma ion, pa e ns and ela ionships o a eal- ime decision making.
A. Knowledge and insigh managemen
Comp ising knowledge based on da a o di e en ypes and o igins, insigh engines
p o ide ele an eal- ime in o ma ion o espond o cu en ends, gaps and
challenges. They can sea ch h ough he in o ma ion o check hei egula o y
compliance and de ec possible anomalies. Insigh s om scien ific, echnical, legal and
ma ke da a may e icien ly ad ance a company's R&D ( esea ch and de elopmen ).
When all he ele an da a is o ganized and app op ia ely dis ibu ed h oughou he
company, he decision-make s a e able o collabo a e and de elop s a egies based on
he sha ed knowledge. On he o he hand, insigh engines’ secu i y measu es include
con olling access and enc yp ing sensi i e supply chain in o ma ion.
B. Cus ome engagemen
Insigh engines can help companies mee he needs and possibili ies o hei cus ome s
in a p oac i e manne . Analyzing ex ensi e da a on cus ome s’ cha ac e is ics and
beha io pa e ns, engines can p o ide eal- ime solu ions o deli e ing pe sonalized
con en a he igh momen . Tools powe ed by insigh engines may imp o e he
expe ience o cus ome s, allowing o e icien sel -se ice and as e o de ulfillmen .
C. In en o y managemen
Apa om p o iding da a secu i y, insigh engines can p ocess di e se in o ma ion in
eal ime and de ec dis up ions a di e en s ages o he supply chain. Toge he wi h
insigh s on he ma ke fluc ua ions and cus ome s’ ac i i y pa e ns, his can help he
company o upda e i s s a egies, op imize s ock o ganiza ion, minimize lead ime and
quickly adap o encoun e ed di icul ies.
D. Ope a ions op imisa ion
Combined da a o a company’s ope a ions, i s consume s, compe i o s and he ma ke
may become a ounda ion o majo s a egies as well as specific p oblem-sol ing.
Ine iciencies and oppo uni ies can be de ec ed and analyzed o c ea e solu ions o
s eamlined p ocesses. Op imizing esou ce use no only con ibu es o cos educ ion
and o e all agili y, bu also helps he company imp o e he sus ainabili y o i s p ac ices.
48 | 249
1.2.3.5. Na u al language p ocessing
Na u al language p ocessing (NLP) “enables compu e s o unde s and, in e p e , gene a e, and
ans o m ex . Cu en s a e-o - he-a models, such as OpenAI’s GPT-4 and Google’s Gemini,
a e able o gene a e fluen and cohe en p ose and display high le els o language
unde s anding abili y” (Maslej e al., 2024, p. 85). NLP allows o he communica ion be ween
machines and humans, wi h AI sys ems in e p e ing na u al language inpu and gene a ing da a
accessible o people. I s analy ical capabili ies include classifica ion o symbols, wo ds and
ph ases, bu also iden ifica ion o unde lying con ex , in en o emo ion o he ex ’s au ho .
A. Knowledge and insigh managemen
NLP enables human agen s o easily access knowledge wi h commands and que ies
o mula ed in na u al language. Wi h he help o speech ecogni ion and op ical
cha ac e ecogni ion ools (OCR), NLP helps analyze any kind o e bal da a. I so s
uns uc u ed bi s o in o ma ion in o o ganized amewo ks and can p o ide
isualiza ions o summa ies o complex da ase s. Backed by ML o deep lea ning
algo i hms, NLP ools a e able o iden i y ele an in o ma ion in he da a and he
ela ionships be ween hem. Fo example, named en i y ecogni ion (NER) is a
echnology used o finding specific elemen s in he ex , such as a ge ed opics,
p ope names o alues.
B. Cus ome engagemen
Th ough na u al language p ocessing, i becomes easie o iden i y ac ual and po en ial
cus ome s’ needs. I can analyze no only he lexical cons uc ion o he sea ch ph ases,
e iews o messages, bu also he meaning o he wo ds and ph ases in con ex . Due o
his, NLP helps ecognise he in en ions o use s, which esul s in a mo e accu a e
depic ion o he p oduc s o se ices ha may be o hei in e es . Fu he p ocessing
o he seman ic con en o ex s allows o sen imen analysis. Wi h ML-based
classifica ion, he s a emen s a e ca ego ized by emo ional ea u es and a ibu ed
posi i e, nega i e o neu al alue. Tha , in u n, p o ides mo e comp ehensi e eedback
in o ma ion and con ibu es o dynamic and pe sonalized in e ac ion wi h cus ome s.
Well-conduc ed na u al language analysis also allows he au oma ed suppo sys em o
decide which que ies o calls a e sui able o cha bo se ice and which demand a
human-consul an ’s in e en ion.
C. Ope a ions op imisa ion
Seman ic sea ch is also beneficial o company’s employees. Wi h NLP echnology i
becomes easie o na iga e h ough documen a ion o websi e’s con en . Along wi h
summa y ea u es his can speed up analy ical asks. When i comes o ma ke ing,
49 | 249
na u al language p ocessing ools p o e o be beneficial in e iewing con en and
p o iding sugges ions o op imized eadabili y, one, g amma as well as
SEO-compliance. Finally, NLP enables con en gene a ion. La ge language models
con ain enough da a o he gene a i e AI o p oduce ex s ha in many cases ma ch he
quali y and e ec i eness o human w i ing. Enhancing abili ies o human agen s in
c ea ing and e iewing con en o mee company’s policies and SEO s a egies educes
he need o a cos ly labo o ce and can significan ly imp o e ma ke ing pe o mance.
1.2.3.6. Compu e ision
Compu e ision “allows machines o unde s and images and ideos and c ea e ealis ic isuals
om ex ual p omp s o o he inpu s” (Maslej e al., 2024, p. 96). Backed by ML algo i hms and
deep neu al ne wo ks, compu e ision is used o asks such as acial and objec ecogni ion,
image classifica ion, agging, seman ic segmen a ion and e en image gene a ion.
A. Knowledge and insigh managemen
Th ough his echnology, isual ep esen a ions o a company's esou ces, including
documen a ion and eal-li e p ocesses, can be cap u ed, a ibu ed meaning and
analyzed o imp o ed, s anda dized pe o mance. Connec ed o senso s in s o es o
wa ehouses, compu e ision de ec s pa e ns and anomalies, p o iding a
comp ehensi e iew o a company's ope a ions, highligh ing gaps in e iciency o
secu i y.
B. Cus ome engagemen
Compu e ision enhances cus ome expe ience in new ways, allowing o close and
be e in o med con ac wi h he i ual ep esen a ions o p oduc s. A pe son can ake
o copy a pic u e o a desi ed objec and upload i o a isual sea ch ool. Image-based
que ies can be e icien when no enough ex ual in o ma ion is ini ially p o ided, ei he
a a s o e o in o dina y li e si ua ions. When combined wi h augmen ed eali y
applica ions o ex ensions, compu e ision makes i possible o isualize a p oduc in a
eal-li e en i onmen , e.g. a pe son’s oom, o pe o m a i ual y-on o a piece o
clo hing/equipmen .
C. In en o y managemen
Compu e ision enables cons an ly upda ed moni o ing o he in en o y. I analyzes
s ock le els and p o ides in o ma ion on hei compliance wi h p edic ed demand o
50 | 249
deli e y schedules. When applied o s o e shel es, his echnology can help op imize he
placemen o p oduc s, so ha hey a e exposed in he mos e icien way.
D. Ope a ions op imisa ion
In-s o e ope a ions can be acked o imp o e he shopping expe ience and gene a e
mo e sales. Hea maps and a ic pa e ns show he a eas whe e cus ome s' p esence
and a en ion may be inc eased, which helps wi h planning an op imal o ganiza ion o
s o es. Compu e ision is also used o ack suspicious beha io o aud, bo h online
and o line. Secu i y appliances ange om de ec ion o inco ec scanning, coun e ei
p oduc s, o iden i ying ake e iews. Objec ecogni ion and image classifica ion can
become e ec i e ools o a websi e’s con en mode a ion.
1.2.3.7. Speech ecogni ion
Speech ecogni ion ools “in e p e human speech and ansla e i in o ex o commands.
P ima y applica ions a e sel -se ice and call ou ing o con ac cen e applica ions;
con e ing speech o ex o desk op ex en y, o m filling o oice mail ansc ip ion; and
use in e ace con ol and con en na iga ion o use on mobile de ices, PCs and in-ca
sys ems. Con ol o consume appliances (such as TVs) and oys is also comme cially a ailable
bu no widely used” (Ga ne , 2024e).
A. Knowledge and insigh managemen
Speech ecogni ion, when implemen ed in o any communica ion sys em wi hin he
company, can collec and analyze me ada a abou in e ac ions. Whe he i is a
con e sa ion in ol ing an employee o a cus ome , he sys em is able o dis inguish
pa e ns o ac i i y (pauses, esponse- ime) as well as emo ional ones o di e en
s a emen s. An ad anced ype o his echnology - oice ecogni ion - allows o
iden ifica ion o a pe son aking pa in he alk.
B. Cus ome engagemen
Insigh s in o clien s’ beha io and emo ions may p o e beneficial o op imizing hei
expe ience. A cus ome can dic a e his/he demands o que ies and engage in
na u al-language con e sa ions wi h minimized loss o ime and a en ion. Da a
ga he ed h ough hese in e ac ions enables be e pe sonalisa ion o an o e and
au oma ed, easy- o-scale-up suppo sys em. In pa allel wi h adi ional sea ch engine
que y p ocesses, companies can imp o e hei answe engine op imisa ion (AEO) and
p o ide an al e na i e so wa e o shopping ac i i ies. Speech ecogni ion ools educe
51 | 249
he need o a clien ’s echnical abili ies and e o s o sea ch, o de and eo de desi ed
p oduc s.
C. In en o y managemen
Voice picking le s he employees con ol he supply wi hou ime-consuming manual
inpu s. Th ough a headse , a wa ehouse wo ke can gi e commands and ecei e
guidance o easily na iga e h ough he s ock and be awa e o i s le els and p ecise
loca ions.
D. Ope a ions op imisa ion
Voice commands p o ide an addi ional channel o communica ion, as well as an easy
access o company’s ope a ional da a. Iden ified gaps and slowdowns in in e ac ions
may sugges solu ions o speeding up he wo kflow. Since any eco ded con e sa ion
can be ansc ibed in eal ime, he c ea ion o up- o-da e documen a ion becomes
au oma ed and ull o de ail.
1.2.3.8. Cha bo s
Cha bo s a e an example o “a domain-specific con e sa ional in e ace ha uses an app,
messaging pla o m, social ne wo k o cha solu ion o i s con e sa ions. Cha bo s a y in
sophis ica ion, om simple, decision- ee-based ma ke ing s un s, o implemen a ions buil on
ea u e- ich pla o ms. They a e always na ow in scope. A cha bo can be ex - o oice-based,
o a combina ion o bo h” (Ga ne , 2024b).
A. Knowledge and insigh managemen
Engaging in in e ac ions wi h cus ome s, cha bo s may seamlessly in oduce su eys o
collec eedback. Along wi h ga he ed me ada a, his p o ides eal- ime in o ma ion on
issues and p e e ences, which makes i easie o de elop ele an sales and suppo
s a egies.
B. Cus ome engagemen
The main ad an age o cha bo s lies in hei a ailabili y. They o e non-s op cus ome
assis ance, managing mul iple in e ac ions simul aneously. Because au oma ed
suppo ou uns he capabili ies o employees wo king wi hin limi ed hou s, i enables
easie scaling-up o he ope a ions and, wha is mo e, imp o es cus ome expe ience.
Real- ime analysis o clien s’ inpu and beha io may esul in mo e pe sonalized and
52 | 249

con ex ual o e s. Cha bo s a e able o sugges ins an ollow-ups di ec ed a a po en ial
buye ’s p e e ences and needs.
C. Ope a ions op imisa ion
Success ul implemen a ion o au oma ed se ices and collec ion o ex ensi e cus ome
da a could con ibu e o inc eased lead gene a ion, clien s’ a achmen o he b and
and, as a esul , highe con e sion a es. Cha bo s elease he human employees om
pe o ming epe i i e asks ela ed o cus ome suppo and may lead o op imized
o de p ocesses. As o he asks s ill pe o med manually by a company’s eam, bo s
can be jus as use ul o assis wo ke s in esol ing issues and sea ching h ough he
da a.
1.2.3.9. Edge a ificial in elligence
Edge a ificial in elligence (edge AI) is “ he implemen a ion o a ificial in elligence in an edge
compu ing en i onmen , which allows compu a ions o be done close o whe e da a is ac ually
collec ed, a he han a a cen alized cloud compu ing acili y o an o si e da a cen e . Edge AI
le s de ices make sma e decisions as e , wi hou connec ing o he cloud o o si e da a
cen e s” (Red Ha , 2024).
A. Knowledge and insigh managemen
Edge AI speeds up he con e sion o knowledge in o decision-making. Sma de ices
p ocess upda ed in o ma ion on he spo wi hou he need o sending hem o da a
cen e s, which minimizes la ency o esul ing ope a ions. Keeping he da a wi hin he
equipmen o a local se e , edge AI ensu es p i acy o sensi i e in o ma ion and allows
o immedia e esponses o c i ical issues.
B. Cus ome engagemen
Fas p ocessing o edge de ices imp o es he e iciency o ex ual, isual and oice
sea ches. Senso s ga he in o ma ion on cus ome s’ beha io and enable eal- ime
pe sonalisa ion o o e s.
C. In en o y managemen
Edge AI keeps he in en o y in o ma ion up o da e and in acco dance wi h supply chain
op imisa ion analy ics. Applied in a wa ehouse o a s o e, sma shel es moni o he
53 | 249
p oduc s o hei quan i y, need o es ock and abno mal ac i i ies. This ensu es he
igh le el o supply and p oac i e p e en ion o unnecessa y loss.
D. Ope a ions op imisa ion
Edge compu a ion dis ibu es he impac o ex ensi e p ocessing h oughou sma
de ices. This no only educes la ency bu also cu s he cos s o ans e ing all he da a
o he cloud. I may become easie o a company o scale up, when addi ional
compu ing asks a e pe o med wi hin he ins alled de ices. Wi h he help o edge AI,
in-s o e ope a ions a e also be e secu ed. Radio- equency iden ifica ion (RFID)
secu i y ags p e en he p oduc s om being sec e ly aken h ough he secu i y ga es,
whe eas au oma ed poin s o sale (POS) epo on any mis-scans o icke -swi ching
almos in eal- ime.
1.2.3.10. Robo ic p ocess au oma ion
Robo ic p ocess au oma ion (RPA) can be unde s ood as “ he so wa e o au oma e asks
wi hin business and IT p ocesses ia so wa e sc ip s ha emula e human in e ac ion wi h he
applica ion use in e ace. RPA enables a manual ask o be eco ded o p og ammed in o a
so wa e sc ip , which use s can de elop by p og amming, o by using he RPA pla o m’s
low-code and no-code g aphical use in e aces. This sc ip can hen be deployed and execu ed
in o di e en un imes. The un ime execu able o he deployed sc ip is e e ed o as a bo , o
obo ” (Ga ne , 2024d).
A. Knowledge and insigh managemen
RPA enables au oma ing complex da a- ela ed asks, including hei collec ion om
a ious sou ces, cleansing, ca ego isa ion and compa isons. The echnology can be se
o pe o m hese ac ions egula ly and p o ide upda ed epo s o quick p oblem
sol ing, s a egy planning and decision making. Au oma ed analysis in ol es eal- ime
da a and p edic ions o gene a e solu ions o op imized p icing, ma ke ing campaigns,
new p oduc s, cus ome se ice o ope a ions’ wo kflow.
B. Cus ome engagemen
Implemen ed in o bo s, RPA enables cons an in e ac ion wi h clien s and gene a ing
immedia e esponses o simple que ies. Au oma ing o de acking p o ides cus ome s
wi h upda es on e e y s ep ha leads om he pu chase o comple ed deli e y. The
eedback can be also collec ed au oma ically and e u n p ocedu es - ca ied ou
wi hou he delay. Ga he ed da a on clien s’ expe iences, pas ac i i ies and beha io is
54 | 249
easily ans e ed be ween CRM and ma ke ing sys ems o ele an ly segmen he a ge
g oups and deli e pe sonalized con en o epea ed pu chases.
C. In en o y managemen
Wi h ex ensi e da a on supply chain ac ual and p edic ed dynamics, RPA au oma es
planning and s eamlines communica ion be ween di e en membe s o in en o y and
ulfillmen sys ems. Regula acking and epo ing keeps he ope a ing eams always up
o da e wi h s ock in o ma ion and eady o immedia e esponse o encoun e ed
p oblems and oppo uni ies.
D. Ope a ions op imisa ion
RPA can handle mundane manual ope a ions wi h be e accu acy and in sho e ime. I
can be assigned wi h a ious accoun ing, eco dkeeping and in oice p ocessing asks
and ensu e hei o ganiza ion, compliance and secu i y. RPA sys ems a e used o keep
in o ma ion up o da e, ha monize ope a ions o a company's sec o s and de elop
dynamic solu ions o p icing, o de ulfillmen and ma ke ing s a egies. Along wi h
ex ensi e cus ome da a, RPA handles in o ma ion on employees and au oma ically
con ols wo k o ce dis ibu ion wi h scheduling and pay oll insigh s.
1.2.3.11. Gene a i e AI
Gene a i e AI is a kind o a ificial in elligence echnology based on machine lea ning. I di e s
om o he kinds o AI in i s abili y o c ea e new da a objec s based on i s aining da ase
a he han making p edic ions conce ning co ela ions be ween da a in i s aining da ase .
This echnology le e ages neu al ne wo ks and deep lea ning o gene a e con en , such as ex ,
images, and e en complex da a se s, making i highly e sa ile and inno a i e.
In oduced in 2022, companies a e s ill expe imen ing wi h i s po en ial. A mo e specific
analysis o he impac o gene a i e AI on he e ail sec o is lacking in he li e a u e we
e iewed. Howe e , du ing in e iews, wo kshops, and da a collec ion o use cases, we
confi med ha some companies a e al eady using i o hei ad an age (see case 3.12). Cu en
applica ions include pe sonalized ma ke ing campaigns, dynamic p icing s a egies, and
au oma ed con en c ea ion o websi es and social media.
Gene a i e AI, such as Cha GPT, has he po en ial o enhance ou main a eas o company
ope a ions:
A. Knowledge and Insigh Managemen : Gene a i e AI can p ocess and syn hesize as
amoun s o in o ma ion, p o iding aluable insigh s and acili a ing knowledge sha ing
55 | 249
and c ea ion. Fo example, i can summa ize la ge da ase s and gene a e
comp ehensi e epo s ha help businesses make in o med decisions.
B. Cus ome Engagemen : Gene a i e AI can enhance cus ome ela ionships by o e ing
pe sonalized esponses, handling inqui ies, and p o iding 24/7 suppo . I can simula e
human-like con e sa ions, imp o ing cus ome sa is ac ion and loyal y h ough mo e
in e ac i e and esponsi e communica ion channels.
C. In en o y Managemen : Gene a i e AI can analyze sales da a and ends o p edic
demand, assis ing in balancing supply and op imizing s ock le els. This helps e aile s
main ain app op ia e in en o y le els, educing was e and imp o ing a ailabili y o
p oduc s.
D. Ope a ions Op imiza ion: Gene a i e AI can au oma e ou ine asks, gene a e epo s,
and p o ide decision suppo , helping o minimize cos s and maximize ope a ional
e iciency. I can s eamline wo kflows, op imize esou ce alloca ion, and enhance
o e all p oduc i i y.
Despi e i s p omising po en ial, implemen ing gene a i e AI comes wi h challenges and e hical
conside a ions. These include da a p i acy conce ns, he need o significan compu a ional
esou ces, and po en ial biases in AI-gene a ed con en . Add essing hese challenges is c ucial
o le e aging he ull benefi s o gene a i e AI while ensu ing e hical and esponsible use.
1.2.3.12. Summa y
Table 8 summa izes main uses o each o he desc ibed AI echnologies in ela ion o he alue
chain in e ail companies.
Table 8 Main uses o AI echnologies in 4 e ail a eas
Technology
Main uses in e ail alue chain (Knowledge and insigh managemen ,
Cus ome engagemen , In en o y managemen , Ope a ions op imisa ion)
Machine
Lea ning
Da a classifica ion, analysis and p edic ions, da a secu i y
Cus ome segmen a ion and pe sonalisa ion
Ma ching s ock le els wi h demand o ecas s
Resou ce alloca ion, ma ke ends, aud de ec ion
Deep
lea ning
Con ex -specific da a pa e ns, eal- ime analysis
Dynamic ma ke ing, hype -pe sonalisa ion
Augmen ed
in elligence
Da a moni o ing and explo a ion, p oblem-sol ing
Sel -se ice
In en o y epo s and sugges ions
Employee suppo
56 | 249
wo k o ce. Wo kshop pa icipan s gene ally exp essed a posi i e and calm pe spec i e on he
swi digi al ans o ma ions d i en by he ise o AI. Pa icipan s om Romania highligh ed ha
AI does no aim o eplace o au oma e asks o displace wo ke s om he labo ma ke .
Ins ead, AI enhances hei exis ing ac i i ies:
Du ing ou esea ch wi h some manage s om s o es like Me o, Cash and Ca y, Selg oss, e c.
we ealized ha AI can be an ex ension o he employee, o he employees, so AI will no
eplace hem, i will be an ex ension and also we can help wi h he aining ha I was alking
abou a ew minu es ago. Because he AI can also be used by he employees, no only by he
cus ome . So he cus ome comes, you know, o a s o e and in e ac s wi h an AI, le 's say an
in-s o e AI assis an , o ge answe s like p ice desc ip ion, whe e i is placed in he s o e, he
employee can also do he same hing. And ha way hey can lea n and hey can inc ease hei
knowledge base abou he p oduc s in he s o e. (Wo kshop_Romania_1)
Du ing a wo kshop, a pa icipan sha ed insigh s de i ed om esea ch conduc ed in Romania,
highligh ing he dual ole o a ificial in elligence in cus ome in e ac ions and employee
enhancemen . AI, exemplified by i ual o desk op cha bo s, au oma es cus ome in e ac ions,
he eby s eamlining p ocesses. Fu he mo e, AI enhances he analy ical capabili ies and
knowledge o employees ac oss all expe ience le els. This mul i ace ed applica ion o AI
ga ne ed a posi i e ecep ion among wo kshop pa icipan s, os e ing an op imis ic ou look
owa ds au oma ion's po en ial benefi s.
The heme o au oma ion eme ged om he analysis o su ey ques ionnai es conduc ed in
Cyp us. Responden s belie e ha AI can eplace company employees in he case o epe i i e
and ime-consuming asks. One in h ee s a es ha echnologies such as cha bo s can ake
o e cus ome se ice and suppo asks, and one in ou belie es ha AI echnologies can
elie e employees o asks such as da a analysis and ex gene a ion (Figu e 3). Mo eo e , AI can
enhance employees' asks by eeing hem om epe i i e du ies. Addi ionally, i can c ea e
ecommenda ions o cus ome s, make decisions, and manage in en o y.
63 | 249

Figu e 3 In wha aspec s do AI echnologies eplace asks pe o med by employees?
(% o esponses)
Sou ce: own elabo a ion
64 | 249
Figu e 4 In wha aspec s do AI echnologies enhance he way asks a e pe o med by
employees? (% o esponses)
Sou ce: own elabo a ion
Au oma ion, acco ding o wo kshop and su ey pa icipan s, boils down o au oma ing basic
and epe i i e ac i i ies, pa icula ly in he a eas o in e ac ion and cus ome suppo , as well as
mo e ad anced wo k in he supply chain. Significan ly, he la e a ea is mo e di icul o
au oma e, as i equi es adap ing he da a in as uc u e o business s uc u e o analy ical
capabili ies and ha ing la ge da a se s on which o ain di e en ia ed AI models. A posi i e
a i ude owa ds change p e ailed among he wo kshop pa icipan s, who poin ed in pa icula
o he posi i e ole o au oma ion in filling he gaps caused by he una ailabili y o ce ain ypes
o employees, as well as s eng hening he wo k o ce in e ms o knowledge and da a analysis.
1.3.4. Good Se ice
The wo kshop pa icipan s explo ed no only he echnical aspec s o p edic ion, op imiza ion,
and au oma ion bu also he quali a i e dimensions o a ificial in elligence and i s applica ion
in he e ail sec o . They iewed his as an oppo uni y o enhance se ice quali y, which
in ol es e ec i e communica ion wi h cus ome s and app op ia e cus ome ca e h oughou
he sales p ocess. These aspec s a e in e wined wi h he challenge o os e ing us in
cus ome in e ac ions and cul i a ing a suppo i e en i onmen o MSME cus ome s.
65 | 249
As p esen ed by a wo kshop pa icipan om I aly, he implemen a ion o AI ools in e ail has o
ake in o accoun he issue o us in cus ome ela ionships:
I would like o d aw a en ion o us : o in oduce AI in o he e ail wo ld, we need o build a
ela ionship o us . The e is - and his is measu ed by nume ous s udies - a ce ain epulsion
o inno a ions ha imply a adical change in habi s. Think o a cus ome whe e he p esence
o a human being is eplaced by AI. The ela ionship wi h he cus ome changes. We need o
know how o in oduce a new ela ionship o us in o his wo kflow. This is achie ed by
choosing sys ems ha can be ailo ed o fine- uned o he con ex in which hey ope a e. I
belie e i is essen ial o add ess how he cus ome expe ience will change, how hei a i ude
may be al e ed, and how hei p opensi y o us may be modified. This is a c i ical elemen
ha de elope s and managemen adop ing AI mus add ess. (Wo kshop_I aly_3)
Acco ding o one wo kshop pa icipan , success ully implemen ing a ificial in elligence
in ol es in eg a ing his echnology no jus wi h da a in as uc u es and o ganiza ional
amewo ks, bu also wi h a obus se ice sys em. The co ne s one o his sys em is us
be ween small o medium-sized en e p ises and hei cus ome s. Upholding us necessi a es
he ca e ul and skill ul deploymen o AI, which should be cus omized o sui he con ex and
unique cha ac e is ics o cus ome g oups h oughou he sales p ocess. This poses a
significan challenge in e ms o de eloping company-wide expe ise and managing
in e ac ions wi h ex e nal s akeholde s. I is he e o e impo an o conside cus ome s'
pe spec i es and a i udes owa ds au oma ed cus ome se ice when designing changes
ela ed o he implemen a ion o AI. On he o he hand, once con ex ual ac o s such as he
iden ifica ion and comp ehensi e unde s anding o cus ome g oups ha e been conside ed, he
pa icipan poin s o he condi ions o oppo uni y ha allow o he aluable use o AI
echnology.
Based on us in cus ome ela ionships and he hough ul use o AI echnology, e ec i e
cus ome ca e can be in eg a ed in o he sales p ocess.
Yeah, I hink ha 's whe e a leas his ma ke segmen a ion comes in. I mean, i 's wi h
cha bo s, ha 's o su e, i 's kind o basic, bu i 's p ecisely because o he a e sales ca e,
somehow gene a ing a leas some messages ha go ou now o pu chase o jus : "You
p oduc is s ill in he baske , maybe you s ill wan o buy i ". This is wha many online shops do
now. Fo example, I ha e an accoun wi h many pla o ms ha sell digi al books. I ge hese
eminde emails all he ime: "The e is s ill a p oduc in you shopping baske , maybe you would
like o buy i ". This is no done by anyone, no by some i ed M Smi h, i is done au oma ically.
Sending newsle e s, ha 's all .... should be doing by now. (Wo kshop_Poland_2)
Among he key ac o s o p o iding excellen cus ome se ice a e he judicious use o
cha bo s o i ual assis an s, and he au oma ion o ma ke ing and cus ome ca e. Howe e , as
wo kshop pa icipan s poin ed ou , deli e ing good se ice necessi a es high-quali y
echnologies capable o esponding o cus ome demands and es ablishing an app op ia e ca e
sys em o hem. Cus ome ca e is unde s ood he e by main aining cons an con ac a all
66 | 249
s ages o sales and a e -sales p ocess. The e o e, he ole o skills in enabling he app op ia e
use o AI in e ail is impo an (as discussed in he nex sec ion).
Table 9 AI unc ions needed in ela ion o challenges in he e ailing sec o
AI unc ions needed
In ela ion o challenges in e ail
P edic i e unc ions
• P edic ing demand in supply chains
• Fo ecas ing sales ends
Op imiza ion unc ions
• Op imizing in en o y le els
• Sus ainable use o esou ces and p oduc s
Au oma ion unc ions
• Au oma ing ma ke ing and cus ome ela ionship
• Managing cus ome eedback
• Au oma ing ou ine asks o complemen ing wo k o ce
“Good se ice deli e y”
• Used skil ully, i helps build a ela ionship o us wi h
cus ome s
• P o ides oppo uni ies o holis ic cus ome ca e du ing he
sales p ocess and a e wa ds
Sou ce: own elabo a ion
Looking a his comp ehensi ely, pa icipan s in co-c ea ion wo kshops ac oss selec ed EU
coun ies highligh ed ou key unc ions o AI in e ail. Fi s ly, AI's p edic i e capabili ies enable
be e managemen o supply chains and sales. Secondly, i op imizes ela ed p ocesses.
Thi dly, pa icipan s emphasized he au oma ion o cus ome se ice and specific asks,
add essing wo k o ce gaps o enhancing exis ing s a capabili ies. Fou hly, AI's po en ial
con ibu ion o enhancing cus ome se ice, pa icula ly in cus ome ela ions, was no ed.
Pa icipan s s essed he impo ance o applying AI wi hin he con ex o MSMEs, conside ing
hei unde s anding o hei cus ome s o a oid unde mining us in hese ela ionships. Only
h ough his app oach can au oma ed and echnology-enabled cus ome ca e, such as i ual
assis an s, be e ec i ely implemen ed.
67 | 249
2. THE STATE OF AI ADOPTION IN RETAIL SECTOR
Key akeaways
■ O e he las decade, he wholesale and e ail ade sec o has g own significan ly.
■ The e ail sec o is domina ed by a high numbe o mic o, small and medium-sized
en e p ises (MSMEs), especially in Romania. This s uc u e can hinde compe i i eness in
in e na ional ma ke s due o lowe ope a ional e iciency and challenges in achie ing
economies o scale, significan ly impac ing he digi aliza ion p ocess.
■ The e ail sec o 's digi aliza ion lags behind o he indus ies such as ICT and p o essional
se ices.
■ E-comme ce sales in he e ail sec o a y significan ly be ween coun ies, indica ing
di e en le els o digi al echnology adop ion.
■ Mos MSMEs in EU coun ies do no use AI echnologies, wi h Poland ha ing he highes
pe cen age o fi ms no using AI.
■ In he e ail sec o , AI adop ion is he lowes among he analyzed sec o s. The mos
common AI applica ions in MSMEs a e obo ic p ocess au oma ion (RPA), speech
ecogni ion, and ex mining.
■ The main ba ie s o AI adop ion a e insu icien specialis knowledge, high
implemen a ion cos s, and legal unce ain ies, limi ing he implemen a ion o hese
echnologies in MSMEs and he e ail sec o . Insu icien specialis knowledge and
di icul ies in hi ing skilled employees a e significan ba ie s o adop ing digi al
echnologies in he e ail sec o and MSMEs.
■ While digi al skills a e inc easingly necessa y o implemen ing digi al inno a ions, he
le el o hese skills among employees emains low, pa icula ly in Poland and Romania,
which hampe s digi al echnology adop ion.
2.1. The main cha ac e is ics o he e ail sec o in he
selec ed coun ies
In he las decade, he wholesale and e ail ade sec o has expe ienced g ow h, excep du ing
he COVID-19 pandemic. This g ow h is pa icula ly no iceable in Romania and Poland, which
lead in u no e , while Ge many and I aly ha e allen below he EU a e age, highligh ing
inc easing dispa i ies in g ow h a es ac oss Eu ope.
A key ea u e o his sec o is he high numbe o small and medium-sized en e p ises (SMEs),
especially in Romania, and o a lesse ex en in Ge many. This s uc u e can limi
compe i i eness in in e na ional ma ke s due o lowe ope a ional e iciency and di icul ies in
68 | 249

achie ing economies o scale. Addi ionally, i plays a c ucial ole in he digi iza ion p ocess.
Da a on u no e and sales olume in he wholesale and e ail ade in selec ed coun ies show
in e es ing ends. Romania and Poland ha e he highes u no e among he analyzed
coun ies (Figu e 5). The cha p esen s he annual index o u no e and sales olume in
wholesale and e ail ade in 2023, using 2021 da a as he e e ence poin (index = 100). Bulga ia
leads among EU27 coun ies and shows an inc ease compa ed o he p e ious yea . Ge many
and I aly ha e esul s below he EU a e age, wi h Ge many also expe iencing a decline
compa ed o 2022.
Figu e 6 shows he long- e m end in u no e and sales olume in wholesale and e ail ade
om 2014 o 2023 o he analyzed coun ies. The da a e eal significan g ow h in his sec o ,
pa icula ly in Romania, which had an index o 55.10 in 2014 and eached 128.40 in 2023,
indica ing a dynamic ma ke de elopmen . In con as , I aly, which s a ed om a highe index
le el o 87.70 in 2014, now occupies he las place. The da a sugges a s able and g owing
de elopmen o he e ail and wholesale ma ke . No ably, since 2020, all analyzed coun ies
ha e shown accele a ed g ow h, likely due o economic eco e y ollowing he COVID-19
pandemic.
A no able cha ac e is ic is he e y high numbe o small and medium-sized en e p ises (SMEs)
in he wholesale and e ail ade sec o (Figu es 7, 8, 9). In he analyzed coun ies, o e 97% o
businesses in his sec o a e SMEs. On a e age, one in fi e SMEs ope a es in he e ail ade
sec o , excep in Ge many, whe e he sha e is lowe a 16.8%, and in Romania, whe e i is highe
a 26%. This s uc u e leads o significan ma ke agmen a ion, which can a ec
compe i i eness and g ow h dynamics in hese coun ies.
This agmen a ion can esul in lowe ope a ional e iciency and di icul ies in achie ing
economies o scale, limi ing in e na ional compe i i eness. Addi ionally, a la ge numbe o
SMEs can lead o g ea e ma ke ola ili y and ins abili y, complica ing long- e m s a egic
planning.
Fu he mo e, he high numbe o SMEs may ace challenges in he digi aliza ion p ocess, as
smalle businesses o en ha e limi ed financial and echnological esou ces, making i di icul
o implemen ad anced digi al echnologies. This can impac hei abili y o adap o changing
ma ke condi ions and hei compe i i eness in he digi al e a.
Indica o s o digi aliza ion le els among en e p ises in he EU suppo his. The digi aliza ion o
SMEs in he analyzed coun ies (Poland, Cyp us, Romania, Ge many, I aly) does no significan ly
di e om he EU a e age (24%) o small and medium-sized en e p ises, excep o Romania
(9%) (Figu e 9). Howe e , hese figu es o SMEs a e conside ably lowe han hose o la ge
en e p ises. Among he analyzed g oup, Ge many and Cyp us achie e he highes esul s, wi h
abou one in h ee SMEs cha ac e ized by ad anced use o digi al echnologies. Romania s ands
ou nega i ely in his ega d, bo h among he analyzed coun ies and wi hin he en i e EU.
69 | 249
Figu e 5. Tu no e and Sales Volume in Wholesale and Re ail T ade – Annual Da a (2023)
Sou ce: Eu os a . Index o Tu no e – To al, Uni : Index, 2021=100, Calenda Adjus ed Da a,
Ci cles – Resul om 2022.
Figu e 6 Tu no e and Sales Volume in Wholesale and Re ail T ade – Annual Da a (2014-2023)
70 | 249
Sou ce: Eu os a . Index o Tu no e – To al, Uni : Index, 2021=100, Calenda Adjus ed Da a
Figu e 7 Numbe o Small and Medium-Sized En e p ises in he Re ail Sec o (2022)
71 | 249
Sou ce: Eu os a . Ci cles – Resul om 2021
Figu e 8 Sha e o SMEs in All Re ail En e p ises (2022)
72 | 249
2.3. Use o ad anced digi al echnologies
In 2023, he as majo i y o SMEs did no use any AI echnology. Among small and
medium-sized en e p ises, he highes pe cen ages o fi ms no using AI we e ound in Finland
(92%), Sweden (86%), and Poland (84%) (Figu e 14). In o he EU27 coun ies, hese alues anged
be ween 70% and 80%. Rega dless o company size, Poland is among he coun ies wi h he
highes pe cen age o en e p ises no using AI. La ge en e p ises show highe le els o AI
adop ion, likely due o g ea e financial and o ganiza ional esou ces enabling in es men s in
mode n echnologies.
The e ail sec o anks lowes in e ms o AI usage (Figu e 15). AI adop ion in e ail is
significan ly lowe han in he ICT sec o , whe e he EU27 a e age is a ound 30%. Among he
analyzed coun ies, Ge many has he highes AI usage in e ail (abou 10%), whe e one in en
e ail en e p ises uses a leas one AI echnology. O he coun ies all below he EU a e age,
wi h ew fi ms employing such echnologies. Fu he mo e, all he analyzed coun ies show low
g ow h in AI usage compa ed o 2021 (Figu e 16). O e he pas wo yea s, AI implemen a ion has
inc eased mos apidly in Cyp us (by abou 2 pe cen age poin s). Re ail en e p ises in Romania
ha e minimal AI usage.
When AI is used (bo h in SMEs and e ail companies), i is mos commonly o obo ic p ocess
au oma ion (RPA) based on AI. RPA au oma es ou ine, epe i i e asks and, when suppo ed by
AI, can also make decisions based on da a analysis, pa e n ecogni ion, o na u al language
p ocessing. This makes business p ocesses mo e e icien , accu a e, and as e (Figu es 17 and
18). O he common AI applica ions in SMEs include speech ecogni ion and ex mining. Speech
ecogni ion con e s spoken wo ds in o ex , imp o ing p ocesses such as cus ome se ice
and mee ing ansc ip ion. This inc eases communica ion e iciency, documen a ion accu acy,
and sa es ime and human esou ces. Tex mining in ol es analyzing la ge ex da ase s o
ex ac aluable in o ma ion such as pa e ns, ends, o sen imen s. In SMEs, ex mining is
used o analyze cus ome opinions and moni o he ma ke . In he e ail sec o , machine
lea ning is also equen ly used, enhancing a ious business aspec s. Fo example, analyzing
cus ome beha io da a allows o pe sonalized p oduc ecommenda ions.
The main ba ie s o AI adop ion in SMEs and he e ail sec o a e insu icien specialis
knowledge (Figu e 19). Ano he significan ba ie is he pe cei ed high implemen a ion cos s o
hese echnologies. Less equen ly epo ed issues include incompa ibili y wi h exis ing
sys ems and unclea legal consequences. On a e age, ac oss all EU coun ies, he bigges
ba ie is insu icien specialis knowledge (Figu e 20). Legal unce ain ies and e hical
conside a ions a e also significan .
Simila di e ences a e seen in he use o o he echnologies, such as cloud compu ing se ices
(Figu es 22 and 23). Poland and Ge many ha e mode a e cloud usage le els, bo h a 44.5%,
while Cyp us and I aly ha e highe alues, a 52.4% and 58.4%, espec i ely. Romania is a he
bo om wi h he lowes le el (13.8%), indica ing significan digi aliza ion delays. In e ms o
79 | 249

pu chasing o ice so wa e, secu i y, and file s o age, I aly leads, while Romania sco es he
lowes , highligh ing he di e ences in digi al echnology adop ion in he e ail sec o among
hese coun ies.
The coun ies also di e significan ly in in eg a ing IoT echnologies (Figu es 23 and 24). The
highes IoT adop ion in he e ail sec o is in Cyp us (40%), while Poland (11%) and Romania (13%)
a e a behind he EU a e age (26%). Cyp us excels in building secu i y (39.1%) and ene gy
managemen (13.9%). Ge many is close o he EU a e age in building secu i y (26.9%), while
Poland and Romania show lowe engagemen in mos analyzed IoT applica ions.
Figu e 14 En e p ises no using AI sys ems (2023)
Sou ce: Eu os a . A ificial In elligence by Class Size o En e p ise (En e p ises do no use any AI
echnologies). The b eakdown includes small and medium-sized en e p ises ( om 10 o 249
employees), all ( om 10 employees), and la ge (250 employees and mo e).
80 | 249
Figu e 15 En e p ises om di e en sec o s ha use a leas one AI sys em (2023).
Sou ce: Eu os a . A ificial In elligence by NACE Re . 2 ac i i y (En e p ises use a leas one o
he AI echnologies).
81 | 249
Figu e 16 Re ail sec o en e p ises using a leas one AI sys em (2023)
Sou ce: Eu os a . A ificial In elligence by NACE Re . 2 ac i i y (En e p ises use a leas one o
he AI echnologies). Ci cles - esul in 2021, ba s - esul in 2023.
82 | 249
Figu e 17 AI applica ions in small and medium-sized en e p ises (2023)
Sou ce: Eu os a . A ificial In elligence by Class Size o En e p ise.
83 | 249
Figu e 18 AI applica ions in e ail ade en e p ises (2023)
Sou ce: Eu os a . A ificial In elligence by NACE Re . 2 ac i i y.
84 | 249

Figu e 19 Reasons o no using AI in SMEs and e ail ade en e p ises (2023)
Sou ce: Eu os a . A ificial In elligence by NACE Re . 2 ac i i y and A ificial In elligence by
Class Size o En e p ise.
85 | 249
Figu e 20 Reasons o no using AI in e ail ade en e p ises (2023)
Sou ce: Eu os a . A ificial In elligence by NACE Re . 2 ac i i y
86 | 249
Figu e 21 Sou ces o da a analysis o e ail ade en e p ises (2023)
Sou ce: Eu os a . Da a analy ics by NACE Re . 2 ac i i y
87 | 249
Figu e 22 Use o cloud compu ing in e ail ade en e p ises (2023)
Sou ce: Eu os a . Cloud compu ing se ices by NACE Re . 2 ac i i y
Ci cles: esul om 2021
88 | 249
Figu e 28 Use o AI and ad anced digi al skills o employees in e ail ade en e p ises (2023)
Sou ce: Eu os a Digi al Skills & En e p ises using a leas one AI echnology.
Figu e 29 Use o AI and p og amming skills o employees (2023)
Sou ce: Eu os a Compu e skills (Indi iduals who ha e w i en code in a p og amming language)
& En e p ises using a leas one AI echnology.
95 | 249

Figu e 30 Companies whe e digi al skills a e becoming inc easingly impo an - SMEs and
e ail ade en e p ises (2023).
Sou ce: Flash Eu oba ome e 529 To wha ex en a e he ollowing skills becoming mo e o less
impo an o you company? "Digi al skills" (e.g., skills equi ed o implemen ing o using digi al
echnologies)
96 | 249
Figu e 31 Employees wi h ad anced digi al skills in he e ail and wholesale ade sec o ,
compa ed o all employees (2023)
Sou ce: Eu os a . (Digi al Skills)
97 | 249
Figu e 32 Employees in he e ail and wholesale ade sec o wi h ad anced digi al skills
(2023). Sou ce: Eu os a . (Digi al Skills). Ci cles - Skills in 2021, ba s - Skills in 2023.
98 | 249
Figu e 33 Sha e o employees wi h ad anced skills in specific g oups o digi al compe encies
(2023)
Sou ce: Eu os a . Indi iduals' le el o digi al skills
99 | 249
Figu e 34 Companies acing di icul ies in hi ing employees wi h he igh compe encies -
SMEs and e ail ade en e p ises (2023).
Sou ce: Flash Eu oba ome e 529 How di icul a e he ollowing asks o you company?
Finding employees wi h he igh compe encies
100 | 249

Figu e 35 Companies acing di icul ies in hi ing specialis s, junio specialis s, and
echnicians - SMEs and e ail ade en e p ises (2023).
Sou ce: Flash Eu oba ome e 529 Does you company ha e di icul ies ec ui ing employees o
he ollowing posi ions? Specialis s, junio specialis s, and echnicians
101 | 249
Figu e 36 Companies whe e he lack o digi al skills among employees hinde s he
implemen a ion o digi al echnologies (2023).
Sou ce: Flash Eu oba ome e 529 Lack o skills hinde s he implemen a ion o use o digi al
echnologies in you company
2.5. Cha ac e is ics o selec ed s a es
2.5.1. Cyp us
Cyp us s ands ou o i s high le el o digi al echnology adop ion, pa icula ly in cloud
compu ing (52.4%) and IoT (40%). While digi al skills a e highly alued, he ac ual le el o hese
skills among employees emains low. The Cyp io e ail sec o uses ad anced echnologies o
building secu i y and ene gy managemen , demons a ing a p oac i e app oach o in eg a ing
new echnologies.
2.5.2. Ge many
Ge many has a mode a e le el o cloud compu ing (44.5%) and IoT (26.9%) usage bu he highes
sha e o e-comme ce in he e ail sec o (20.9%). Ge man fi ms show be e ad ancemen in
102 | 249
digi al echnology use bu s ill ace issues wi h he sho age o employees wi h abo e-basic
digi al skills. Despi e his, Ge many main ains a s ong posi ion in he digi aliza ion o he e ail
sec o compa ed o o he coun ies.
2.5.3. I aly
I aly ea u es he highes le el o cloud compu ing usage (58.4%) in he e ail sec o and a high
le el o e-comme ce adop ion. Howe e , I alian SMEs s uggle wi h a lack o specialized digi al
knowledge, hinde ing he implemen a ion o new echnologies. I aly holds a e age posi ions in
AI and IoT usage, indica ing he need o u he digi aliza ion in es men s o enhance ma ke
compe i i eness in Eu ope.
2.5.4. Poland
In Poland, he wholesale and e ail ade sec o epo s high u no e bu is cha ac e ized by a
low le el o digi aliza ion, especially among small and medium-sized en e p ises (SMEs). In
2023, only 7.7% o e ail companies used e-comme ce channels, and 84% o SMEs did no use AI
echnologies, indica ing a need o in ensified digi aliza ion e o s. Addi ionally, Polish fi ms
s uggle o hi e employees wi h he necessa y digi al skills, which hinde s he implemen a ion
o new echnologies and lowe s ma ke compe i i eness.
2.5.5. Romania
Romania eco ds he highes g ow h in he wholesale and e ail ade sec o bu also has he
lowes le el o digi aliza ion among he analyzed coun ies. Only 13.4% o e ail companies use
e-comme ce, and 13% use IoT, indica ing significan delays in adop ing digi al echnologies.
Addi ionally, Romanian fi ms ace significan challenges in hi ing employees wi h he necessa y
digi al skills, limi ing hei adap abili y and ma ke compe i i eness.
103 | 249
3. REAL-LIFE AI USES IN RETAIL COMPANIES: USE CASES
AND BEST PRACTICES
In his sec ion, we p esen how selec ed e ail companies inco po a e a ificial in elligence
echnologies in o hei e e yday ope a ions. We compiled hese use cases and bes p ac ices
h ough a e iew o academic and g ay li e a u e, consul a ions wi h indus y expe s, and
pa icipa ion in indus y con e ences. The desc ip ions a e based on he in o ma ion sou ced
h ough each company’s websi e and, in se e al cases, h ough in e iews wi h he
company’s ep esen a i es.
Key akeaways
■ In he majo i y o cases, AI was specifically used o op imizing cus ome suppo , mos ly
h ough in e ac ions wi h cha bo s and au oma ed pe sonaliza ion o o e s.
■ O he uses ocused on au oma ed SEO p ocesses, supply chain managemen , con en
gene a ion and mode a ion o gene al da a o ganiza ion and analy ics.
■ Ou sou ced AI ools a e o en o e ed as a se ice, he e o e ma ke ed as equi ing no
echnical skills o main enance e o s.
■ Ne e heless, se e al cases sugges a need o basic unde s anding o he company’s
ope a ional da a, as well as he knowledge on AI ool’s unc ionali y and he benefi s o i .
■ Deepe insigh in o he implemen a ion p ocess e ealed a need o c i ical hinking skills
o e alua ing ools’ pe o mance and communica ion skills o sha ing he collec ed
insigh s.
■ The analyzed cases end o po ay au oma ion o manual p ocesses as a posi i e change,
ocusing ei he on human-machine coope a ion o he change o employees’ ocus om
epe i i e o mo e complex asks.
To be no ed, while we e e o “use cases” o highligh he di e si y o applica ions o di e en AI
echnologies, he esea ch p ocess e ealed ha in business p ac ice he e m is o e used o
ma ke ing pu poses. Fo echnology p o ide s, “use case” usually means a gene al desc ip ion
o any possible use o an AI ool. This, howe e , begs he ques ion o how applicable, in eali y, a
gi en solu ion would be o an MSME wi h limi ed esou ces. Tha is why we ocused on he da a
p o ided in mo e specific cases.
F equen ly named “case s udies” o “cus ome s o ies”, hese accoun s desc ibed success ul AI
applica ions in ac ual companies wi h p ecise in o ma ion on he implemen a ion’s esul s. This
could occu in he o m o KPIs (e.g. con e sion a e, cus ome sa is ac ion sco e) o an
explana ion by a company ep esen a i e o he impac on he ope a ions/employees. While
using hese measu es may indica e a well-g ounded success o he desc ibed AI uses, i should
be emembe ed ha he me ics could ha e been defined wi h he in ol emen o he
echnology p o ide s, p esumably in a o o hei p oduc ’s e iciency.
The companies in which AI was implemen ed we e all ca ego ized as MSMEs, using he numbe
104 | 249
ⓘ
Shipup is an o de moni o ing and communica ion ool ha allows o acking o
shipping p ocesses and p o iding eal- ime ale s o con ac cus ome s be o e hey
expe ience incon eniences o delayed o de .
Th ough Shipup, Feed. could au oma ically in o m clien s abou o de issues. This, in u n,
educed he numbe o icke s ecei ed by he suppo eam and enabled quick sol ing o
encoun e ed di icul ies.
Addi ional use o Magen o10 and Diduenjoy ools11 helped he company collec mo e aluable
in o ma ion on hei cus ome s. The fi s so wa e was used o ob ain o de de ails, while he
o he one became a sou ce o eedback and cus ome expe ience da a.
ⓘ
Magen o O de Managemen Sys em (OMS) is a solu ion o moni o ing and managing
in en o y, ensu ing i s isibili y on di e en s ages o supply chain and esponsi e
cus ome se ice.
ⓘ
Diduenjoy o e s solu ions o op imized cus ome expe ience h ough ac ionable
eedback. I enables su ey c ea ion and mul ichannel dis ibu ion, as well as o ganized
eedback collec ion om di e en sou ces.
Ex ensi e cus ome expe ience da a p o ed o be help ul no only o he s eamlined suppo
sys em. E ec i e communica ion wi hin he company esul ed in ans e ing aluable insigh s
ha influenced u u e ma ke ing s a egies, newsle e w i ing and e en p oduc design.
3.2.4. Resul s
The implemen ed solu ions led o a significan inc ease in Feed.’s sales. In less han a yea , he
li e cha con e sion a e eached 6%, which esul ed in €180,000 e enue. The icke
managemen was op imized, eleasing he employees om excessi e wo kload. Ticke s ha
equi ed mul iple ouchpoin s (employee’s ac ions), once adding up o 80% o all icke s, we e
now educed o 10% and o a g ea ex en eplaced by one- ouch icke s. Op imized cus ome
suppo e en ually p o ed o ealize he idea o genuine “ oyal y expe ience” wi h Cus ome
Sa is ac ion (CSAT) sco es jumping om 75 o 94%.
3.2.5. Employee skills needed o adop he solu ions
Despi e limi ed esou ces, Feed. was able o easily adop new echnologies and make use o
hem h ough he eam o 6. No de elope skills o addi ional suppo we e equi ed o benefi
om Zendesk’s sys em. Analy ics ools we e equipped wi h da a isualiza ion and epo ing
ea u es and he company only needed a p ope communica ion sys em o sha e he acqui ed
knowledge o in o med decision making.
11 h ps://www.diduenjoy.com/en/in eg a ion/zendesk
10 h ps://www.zendesk.com/ma ke place/apps/suppo /321722/magen o-2-by-zenpla e
111 | 249

3.3. G ünewald
3.3.1. Desc ip ion o he company
Sou ce:
h ps:// ood o ecas .com/en/ e e ences/he -g uenewald- on-de -baecke ei-g uenewald/
Name
Coun y
Size
G ünewald
Ge many
<250 employees
Desc ip ion: G ünewald is a Waldlaube sheim-based bake y ha has exis ed in he Ge man
ma ke o o e 125 yea s. Today, wi h 29 b anches in he egion, he company p ese es
adi ion while emb acing changes o he mode n wo ld. I p o ides an example how new
echnologies can wo k in a o o sus ainabili y, keeping he business sa e o he u u e
gene a ions.
Goal: Op imizing o de ing p ocesses o be e ma ch he demand o specific i ems.
S a egy: Au oma ed o de ing sys em based on egula o ecas s and cen ally con olled.
Technology p o ide : ood o ecas
Tools: ML- and DL-powe ed so wa e o o de au oma ion.
3.3.2. The need
As a company wi h long adi ions, G ünewald made conside able e o s o keep he business
local. In spi e o scaling up, he bake y p ocu es aw ma e ials om supplie s wi hin he egion.
Ca ing o he en i onmen and financial sus ainabili y o he business, G ünewald wan ed o
use hei esou ces p uden ly, while main aining he es ablished quali y. Wi h he ene gy and
aw ma e ials p ices g owing, op imal supply chain managemen was indeed a ma e o
su i al.
112 | 249
Ha ing de eloped a sizable chain o s o es, howe e , he company’s managemen ealized ha
some o ganiza ional p ac ices we e no fi o he new business model. To da e, indi idual
selle s ulfilled o de s by hemsel es based on hei in ui ion and expe ience. E en i he e we e
ins uc ions om abo e as o which p oduc s in wha quan i ies should be o de ed, he decision
was e en ually de e mined by a wo ke ’s, some be e some wo se, pe sonal judgmen . Wi h he
numbe o b anches inc easing, new selle s had o be hi ed. This made he o de ing p ocess
e en mo e p one o e o s.
As a esul , he in en o y p epa edness a ied in ime and place. On he one hand, unexpec edly
high e u ns o unsold p oduc s occu ed, while in di e en cases some p oduc s we e ou o
s ock oo soon. Las ly, manual o de gene a ion engaged employees’ ime and blocked he cash
egis e s used in he p ocess.
In his pa icula case, i was he echnology p o ide , ood o ecas 12, ha eached he po en ial
cus ome :
ⓘ
Food o ecas ( o me ly we ks a. ) is a Ge man company ha “uses a ificial
in elligence (AI) o educe ood was e” h ough hei So wa e as a Se ice (SaaS) model.
I combines a gi en company’s his o ical da a wi h ex e nal ac o s o p edic u u e
demand and au oma ically op imize p oduc ion, o de ing and s o e planning.
The main goal o G ünewald was o efine hei o de ing sys em wi h au oma ion, educing he
consequences o human e o s and o e coming demand unce ain y.
3.3.3. AI-based solu ion
Con incing au oma ion
G ünewald’s POS da a was analyzed wi h ood o ecas ’s so wa e in o de o implemen he
O de op imisa ion solu ion13:
ⓘ
AI-based O de op imiza ion is one o he ools buil on ood o ecas ’s algo i hm ha
p o ides daily o ecas s a o de i em le el. The p edic ions o sales olumes a e
au oma ically ansla ed in o o de p oposals, eady o be igge ed o manually
adjus ed by employees.
13 h ps:// ood o ecas .com/en/unse e-loesungen/
12 h ps:// ood o ecas .com/en/
113 | 249
A e success ul es ing in one o he G ünewald’s s o es, he solu ion was in oduced in o he
loca ions. Each day, he sys em gene a ed o de s based on he o ecas o he ollowing day’s
sales o di e en p oduc s. Ini ially, he mo e expe ienced employees we e skep ical and
modified he au oma ic o de s. As he ime passed, he p edic ions made by AI so wa e u ned
ou o be so accu a e ha he change was uni o mly emb aced. E en ually, he en i e o de ing
p ocess was au oma ed. The employees could s op bo he ing wi h manual calcula ions and
hei unce ain p edic ions and ully concen a e on selling he p oduc s.
Mo e con ol wi h less e o
All o he da a became o ganized and easily a ailable o he managemen o o e look he
ope a ions and se s a egic a ge s, e.g. conce ning he desi ed balance be ween e u ns and
s ockou s, al hough as he sys em p o ed success ul and consis en , he op-down adjus men s
could also become less equen . T anspa ency o in o ma ion enabled unp eceden ed insigh
in o disc epancies and became a eliable basis o he p e en ion.
3.3.4. Resul s
The implemen ed solu ions led G ünewald o be e financial ou comes. The e u ns became
smalle in size, and demanded p oduc s we e sold in la ge quan i ies, in some cases eaching
80% inc ease in sales. Wi h educed loss o esou ces, he company had s ong e idence o
claim sus ainabili y in on o i s clien s.
3.3.5. Employee skills needed o adop he solu ions
Since he success o he AI sys em was based on lowe le els o employee in e en ion, i may
sugges li le o no need o addi ional skills. G ünewald’s manage also claimed ha in case o
oubles, ood o ecas ’s suppo was always a ailable. Howe e , he hen s a ed ha he
echnology p o ide could “ educe he complexi y e en u he ”. Con incing skep ical
employees owa ds he sys em was said o be enabled mos ly due o echnology’s e iciency bu
he me e occu ence o he p oblem may poin o he need o adap a ion and change
managemen skills.
3.4. Ke igans
114 | 249
3.4.1. Desc ip ion o he company
Sou ce:
h ps://www.shopbox.ai/cases udies/shopbox-ai- ans o ms- he-shopping-expe ience- o -c
a -bu che -ke igans-boos ing-con e sions-by-250
Name
Coun y
Size
Ke igans
I eland
<50 employees
Desc ip ion: Ke igans is a amily- un c a bu che . Along wi h 5 loca ions in no he n Dublin,
he company conduc s i s sales online, deli e ing p oduc s all o e I eland. The o e anges
om adi ional cu s o eady- o-ea low- a meals, along wi h ba becue and cooking
condimen s. Owne s ake p ide in he quali y o he mea s, assu ing a ully anspa en
a m- o- o k supply chain.
Goal: Engaging and p ofi able e-comme ce wi h limi ed esou ces.
S a egy: Pe sonalized cus ome expe ience h ough au oma ion ools.
Technology p o ide : Shopbox AI
Tools: AI Cu a ed Homepage, AI Shop Assis an
3.4.2. The need
B endan Ke igan alued locali y and di ec con ac wi h he cus ome s since he ounded his
amily business in 1973. In he p esen , he company ope a ed by his wo sons expanded i s
ope a ions o he ecomme ce model.
Keeping up wi h mode n equi emen s o he e ail indus y could no , howe e , be limi ed o
launching an online s o e wi hin he Shopi y pla o m. The owne s wan ed o p ese e he
quali y o se ice p o ided by hei small eam in he physical loca ions. This adi ional
app oach was also mo i a ed by he company’s limi ed esou ces.
Ke igans en e ed in o pa ne ships wi h alliance g oups o na ional e aile s ha , o a poin ,
helped hem become ai compe i ion o la ge companies. Ye , o he new ecomme ce channel
o se e i s pu pose in an e icien way, some change in ope a ions managemen was
necessa y. Ke igans needed mo e online p esence and con e sions wi hou o e ly exploi ing
i s s a and finances.
115 | 249
The solu ion o his challenge eme ged when he company came ac oss Shopbox AI14 se ices.
ⓘ
Shopbox AI is a sales engine o op imizing adi ional cus ome expe ience in
ecomme ce en i onmen s. Fully au oma ed and main enance ee, he so wa e
gene a es con e sions h ough pe sonalisa ion ools, in elligen me chandising and
unob usi e assis ance s a ing om he use ’s fi s click wi hin he websi e.
The main goal o Ke igans was o main ain he in-s o e quali y o i s cus ome se ice, while
engaging mo e online cus ome s o pu chase he p oduc s.
3.4.3. AI-based solu ion
Pe sonal websi e
A e only 3 days o p epa a ions, Shopbox AI implemen ed hei fi s echnology in o Ke igans’
Shopi y s o e - Cu a ed Homepage15.
ⓘ
Cu a ed Homepage is a ea u e ha pe o ms dynamic pe sonalisa ion based on a
use ’s e e y in e ac ion wi h a websi e. Th ough a ge ed layou and p oduc o e s i
p o ides a unique shopping expe ience o each cus ome .
Cu a ed Homepage displayed ecommenda ions, p omo ions and ecen ly iewed o ending
i ems. As a esul , new clien s could become engaged om he e y fi s isi o he online
s o e, whe eas p e ious use s gained quick access o he p oduc s hey we e in e es ed in.
Recommended i ems shown in p oduc display page (PDP) ca ousels no only ollowed
cus ome s’ p e e ences bu we e also linked o Ke igans’ in en o y and ma ke ing da a. S ock
le els, ep icing and p omo ions we e au oma ically in eg a ed wi h he ac ual o e . This way,
he p ocess equi ed no in e en ion on he pa o he company’s small eam.
Main aining excellen se ice
As Ke igans’ o e is designed o ou ma ch he egula supe ma ke supplies’ quali y, i mus
also come wi h he p ope cus ome suppo . In physical s o es, employees made su e ha he
clien s we e se ed wi h ca e o hei indi idual needs.
To ma ch he le el o online shopping expe ience wi h adi ional in-s o e encoun e s, Shopbox
AI in oduced i s AI Shop Assis an 16.
16 h ps://www.shopbox.ai/blog/how-ai-shop-assis an s-a e-changing- he- e ail-indus y
15 h ps://www.shopbox.ai/shopbox-demo
14 h ps://www.shopbox.ai/
116 | 249

ⓘ
AI Shop Assis an is a cus ome se ice ool p o iding pe sonalized ollow-ups o
simila p oduc s, up-selling and c oss-selling. Th ough ailo ed o e s i helps inc ease
he a e age o de alue (AOV) and educe he bounce a e.
AI Assis an enabled Ke igans o p esen i s wide ange o p oduc s wi hou b inging he
clien s o one o hei loca ions. The ully au oma ed p ocess no only educed he demand o
employees’ cus ome suppo bu also did no equi e any echnical main enance. The
dis inc i e quali y o his app oach was ha i allowed o ad anced cus ome expe ience
wi hou he need o subs an ial amoun s o his o ical da a ha a e usually used o
pe sonalisa ion.
3.4.4. Resul s
The implemen ed solu ions led he small company o each g ow h. Shopbox ools, while s ill
engaging only 39% o he online cus ome s, now media e 59% o ansac ions. Spending wice
as much ime on Ke igans’ websi e and use s iew h ee imes mo e p oduc s. This ansla ed
o a 250% inc ease in con e sions and an a e age o de alue highe by 9%.
3.4.5. Employee skills needed o adop he solu ions
Since he solu ions o e ed by Shopbox AI a e based on ull au oma ion, he echnology p o ide
claims ha no IT suppo o main enance is equi ed o benefi om i s se ices.
3.5. Kuchyne Valen
3.5.1. Desc ip ion o he company
Sou ce: h ps://pl.sem ush.com/company/s o ies/kuchyne alen /
Name
Coun y
Size
Kuchyne Valen
Slo akia
<50 employees
117 | 249
Desc ip ion: Kuchyne Valen is a amily- un ki chen and u ni u e manu ac u ing company
based in B a isla a. KV o e s hei p oduc s o indi idual cus ome s h ough he company's
websi e h ps://www.kuchyne alen .sk/ which combines an online s o e and a blog.
Goal: “C ea ing excep ional blog con en and ensu ing i s widesp ead dis ibu ion.”
S a egy: SEO, image op imiza ion
Technology p o ide : Sem ush
Tools: Si e Audi , Keywo d Magic Tool, Keywo d Gap, Keywo d O e iew
3.5.2. The need
A e expanding hei sales channels o he digi al en i onmen , Kuchyne Valen aced new
challenges o eaching he online cus ome s. Wi h limi ed esou ces, and li le SEO expe ience,
hey se on a ask o de elop a success ul ma ke ing s a egy ha would make he company
isible and i s o e - ma ching po en ial clien s’ sea ches.
The company’s websi e se es wo pu poses. I s online s o e p o ides a sales channel
posi ioning i in an e-comme ce model. The blog, on he o he hand, ac s as a ma ke ing ool
helping Kuchyne Valen showcase i s solu ions and finished p oduc s. I also became he main
sou ce o sha ed con en con ibu ing o isibili y o he company's o e .
Ini ially, he eam unning KV made independen e o s o p omo e hei company using
di e en SEO echniques. The esul s, howe e , p o ed o be unsa is ac o y and Kuchyne Valen
eached ou o analy ic ools o e ed by Sem ush17:
ⓘ
Sem ush is he online isibili y managemen and con en ma ke ing SaaS pla o m. I
p o ides se ices such as websi e audi ing, keywo d esea ch, backlink op imiza ion,
compe i i e analysis, con en c ea ion and anking moni o ing.
The main goal o he company was o imp o e hei blog con en so ha i gene a ed mo e leads
which, in e ec , would inc ease sales and company’s e enue.
17 h ps://www.sem ush.com/
118 | 249
3.5.3. AI-based solu ion
Unde s anding he p oblem
Fi s s ep aken by he Sem ush eam was o unde s and he p oblems ha Kuchyne Valen was
acing and find a solu ion ailo ed o hei specific needs. Using he Sem ush Si e Audi 18, he
eam iden ified specific influence o a ious SEO echniques on he company's o ganic a ic.
The audi p o ed ha he o me s a egy, which a o ed copyw i ing, could be imp o ed by an
app oach ha pu mo e emphasis on image op imiza ion.
ⓘ
Si e Audi - SEO analysis ool, c awling and checking e e y page on a gi en websi e. I
p oduces egula epo s on websi e’s issues, hei le els o p io i y and solu ions o fix
hem.
Op imizing con en wi h keywo ds
Well-selec ed keywo ds helped Kuchyne Valen o ganize hei web con en in acco dance wi h
SEO demands. Wi h he help o Sem ush Keywo d Magic Tool19, hey we e able o find he mos
e icien keywo ds o image i les. All he keywo ds we e ca ego ized, c ea ing clus e s wi h
ele an subca ego ies and ela ed sea ches. Tha helped define he s uc u e o he whole blog
con en and dis ibu e images ac oss di e en ca ego ies and subca ego ies.
ⓘ
Keywo d Magic Tool - opic-based keywo ds sugges ions o op imized and s uc u ed
con en o ganiza ion. Shows possible links be ween mo e gene al opics, subg oups
and ela ed sea ches, including long- ail keywo ds.
Unde s anding compe i ion
Using he Sem ush Keywo d Gap20, KV could analyze hei compe i o s’ bes -pe o ming
keywo ds ha we e no as e icien ly used by he company. Iden ified gaps and s eng hs ac ed
as a basis o a ge keywo d mas e lis .
ⓘ
Keywo d Gap - enables a compa ison o keywo d p ofiles be ween compe ing
companies. Lis s all common and unique keywo ds (including o ganic, paid and PLA) a
gi en fi m anks o wi h alues and c oss-company di e ences and in e sec ions.
Unde s anding ends
Keywo d O e iew21 helped KV iden i y ending keywo ds which esul ed in de eloping a
specific p oduc line ha me he demand, inc eased blog’s SERP sco e and le he company
achie e expe s a us in he specific a ea.
21 h ps://www.sem ush.com/analy ics/keywo do e iew/
20 h ps://www.sem ush.com/analy ics/keywo dgap/
19 h ps://www.sem ush.com/analy ics/keywo dmagic/
18 h ps://www.sem ush.com/si eaudi
119 | 249
ⓘ
Keywo d O e iew - keywo d esea ch da abase wi h gene al in o ma ion on a gi en
keywo d, including sea ch olume, di icul y, in en and CPC alue.
3.5.4. Resul s
The implemen ed solu ions helped Kuchyne Valen achie e 250% e enue g ow h. O e he nex
5 yea s, he websi e's o ganic a ic emained he p ima y sou ce o lead gene a ion. La e on,
he company expanded i s in ake o ma ke ing echnologies and s a ed PPC ad e ising using
Sem ush PPC insigh s and Google T anspa ency Cen e .
3.5.5. Employee skills needed o adop he solu ions
O e all, main gaps in KV’s esou ces we e: a) lack o da a, b) lack o unde s anding o o -page
SEO (sea ch engine ankings influenced by p ocesses conduc ed ou side he websi e).
3.6. L Cosme ics
3.6.1. Desc ip ion o he company
Sou ces: h ps://www.leafio.ai/case-s udies/l-cosme ics/, an in e iew wi h Leafio
ep esen a i e
Name
Coun y
Size
L Cosme ics
Es onia
<250 employees
Desc ip ion: L Cosme ics is a d ugs o e e ail chain selling hygiene, heal h and pe sonal ca e
p oduc s o women, men and child en. Thei o e includes o e 90 b ands om global and
domes ic p oduce s. Sales ake place h ough o e 25 s o es loca ed all o e Es onia and an
e-comme ce websi e.
Goal: C ea ing an op imized in en o y managemen sys em.
S a egy: Au oma ion o o de p ocessing and s ock managemen .
120 | 249
dependency on human suppo agen s, he company was unp epa ed o mee he pe iodically
ising demand wi hou he delay o se ice.
Dissa isfied wi h o me p o ide s and modes o ope a ion, P ocosme decided o y ou he
o e o Tidio26, a pla o m p o iding AI-based ools and solu ions o op imized cus ome
se ice.
ⓘ
Tidio - “Ou sui e o li e cha , cha bo s, helpdesk ools, and AI solu ions helps b ands o
all sizes d i e hei businesses o wa d by o e ing quick and quali a i e suppo and
c ea ing eal connec ions wi h hei cus ome s.”
The main goal o he company was o implemen a single app ha would handle all he cus ome
se ice asks and be accessible o he employees and clien s alike. As a esul , new con ac s
would be gene a ed o c ea e “a heal hy communi y” o in e es ed and sa isfied use s.
3.8.3. AI-based solu ion
One app o all
To adjus Tidio’s echnology o hei specific needs, he P ocosme ’s eam decided o choose
Tidio+ plan27 wi h cus omized analy ics and au oma ion solu ions.
ⓘ
Tidio+ p o ides a se o ailo ed cus ome se ices, including dedica ed suppo eam,
li e cha , cha bo s and icke ing sys em. I allows he companies o imp o e he a es o
hei leads, con e sions, sales and cus ome sa is ac ion wi h quick e u n on
in es men and easily scalable solu ions.
Wi h Tidio+, he company was able o ca y ou all he necessa y asks wi hin one en i onmen .
O de managemen , da a collec ion and cus ome suppo could now be conduc ed in an
e icien and consis en manne .
Consolida ing communica ion h ough li e cha
Li e cha ea u e28 enabled employees o gain easy access o o de in o ma ion. Any
modifica ions, cancella ions and e unds can be quickly p ocessed h ough he con e sa ion
panel. S eamlining communica ion h ough one ool sped up he eac ion ime and unified
con e sa ional da a o u he analyses.
28 h ps://www. idio.com/li e-cha /
27 h ps://www. idio.com/ idio-plus/
26 h ps://www. idio.com/
127 | 249

ⓘ
Tidio’s li e cha , ailo ed o e-comme ce and small businesses, is a cus ome suppo
ool o managing communica ion channels. I p o ides access o mul iple in e ac ion
pla o ms om any wo kplace, which leads o quicke esponding, inc eased cus ome
engagemen and cons an acking o use s’ ac i i y o be e pe sonalisa ion and
highe sales.
Au oma ed in e ac ion
One o he majo changes ha occu ed h ough implemen a ion o Tidio’s sys em was he
in oduc ion o cha bo s29. Connec ed o he company’s ope a ions hey p o ed use ul in
ca ying ou nume ous asks wi h unp eceden ed e iciency and speed.
ⓘ
Tidio’s cha bo s au oma e sales suppo and cus ome se ice h ough engaging in
in e ac ions wi h clien s and collec ing da a gene a ed in he p ocess. Th ough
pe sonalized assis ance, hey can simpli y lead gene a ion, inc ease a e age o de
alue and educe ca abandonmen . Real- ime mul i-language con e sa ions allow o
e ec i e communica ion u he imp o ed by ad anced p oblem-sol ing and a ge ed
ecommenda ions.
In P ocosme ’s case, he e a e fi e cha bo s se ing a di e se ange o unc ions, such as FAQ
answe ing, newsle e managemen , da a collec ion and cus ome se ice. They a e able o
pe o m hei asks ou side egula wo king hou s and assis many clien s a he same ime.
Th ough iendly in e ac ions, cha bo s g ee use s, o e help and c ea e an in i ing
a mosphe e o keep he cus ome s engaged and inclined o finalize he ansac ions.
Ge ing he da a
Cha in e ac ions allow he po en ial clien s o ge o know he o e ha migh sui hei
pe sonal needs, esol e issues, as well as subsc ibe o he newsle e and gi e hei con ac
in o ma ion. The e iciency o hese p ocesses is bo h enabled by and u he beneficial o
analy ical ools o e ed by Tidio30.
ⓘ
Tidio’s analy ics ools collec and p ocess da a o moni o pe o mance and p o ide
ac ionable insigh s. They keep ack o he cha flows, ac i i y pa e ns and eme ging
issues o op imize ope a ions and imp o e cus ome engagemen .
De ailed analy ics allow o cons an moni o ing o sales da a and de ails o cus ome s’ beha io
and p e e ences. In o ma ion om p e ious ansac ions esul s in deepe unde s anding o
obs acles and oppo uni ies ha could be used o quick p oblem-sol ing and up- o-da e
wo kflow op imisa ion. Real- ime imp o emen s and da a-based suppo we e key o inc eased
sa is ac ion o he cus ome s.
30 h ps://www. idio.com/analy ics/
29 h ps://www. idio.com/cha bo -ai/
128 | 249
3.8.4. Resul s
The implemen ed solu ions led o imp o ed P ocosme ’s KPIs in many a eas. They we e able o
launch a ma ke ing campaign ha esul ed in o e €1000 e u n on in es men . The o e all
con e sion a e was no only s abilized bu sys ema ically inc eased by 27% in a yea . Along wi h
ha he company achie ed a 23% g ow h in sales. Mon hly lead gene a ion changed om 10-30
o o e 100. Email opening a e eached an a e age o 18-22%. Acco ding o P ocosme ’s eam,
Tidio’s solu ion helped gene a e o e ⅓ o hei cu en ecomme ce e enue. The changes we e
also app ecia ed by he clien s. A e age a ings o hei e iews g ew om 3.8 o 4.75 ou o 5
s a s.
3.8.5. Employee skills needed o adop he solu ions
A e ha ing issues wi h he complexi y o p e ious ope a ing sys ems, P ocosme ’s eam
seeked “an easy- o- each app” o se e hei needs. Tidio p omo es hei p oduc s as
accessible, wi hou a need o p og amming ha would equi e IT specialis s o wo k wi h.
3.9. Pu elei
3.9.1. Desc ip ion o he company
Sou ces: h ps://www.ul ima e.ai/cus ome -s o ies/pu elei,
h ps://www.zendesk.de/blog/die-zielg uppe-an-de - ich igen-s elle-abholen/
Name
Coun y
Size
Pu elei
Ge many
<250 employees
Desc ip ion: Founded in 2016, Pu elei o e s jewel y and li es yle p oduc s inspi ed by Hawaii.
The people behind he company ake p ide in he du abili y o ma e ials, collabo a i e design
p ocess and he guiding “Aloha way o li e”. Th ough a ious sales channels and communica ion
media, Mannheim-based Pu elei eaches cus ome s ac oss Eu ope.
Goal: Quali y cus ome expe ience wi hou o e exploi ing he suppo eam.
S a egy: Pe sonalized sel -se ice and au oma ion o epe i i e asks.
Technology p o ide s: Zendesk, Ul ima e
Tools: Zendesk Explo e, Ul ima e Zendesk in eg a ion, Ul ima eGPT
129 | 249
3.9.2. The need
“Be you own cus ome ” - is he mo o o Pu elei’s cus ome se ice eam. Wi h an in e na ional
g oup o clien s and nume ous o ms o communica ion anging om elephone, e-mail o
di e en social media cha s, he company needed an o ganized sys em o suppo .
The idea o p o iding com o and ca e o e e y clien esul ed in a help cen e de eloped
h ough Zendesk pla o m31:
ⓘ
Zendesk is a cus ome se ice so wa e p o iding a comple e CS solu ion ha is bo h
scalable and easy o use. I s ange o ools a e expec ed o imp o e cus ome
expe ience, op imize employee p oduc i i y and p o ide aluable insigh s in o
companies’ ope a ions. Thei se ice is based on o iginal solu ions in eg a ed wi h
ex e nal apps and echnologies.
The help cen e ga he ed da a om a ious communica ion channels and con ained a
knowledge base o FAQs and help ul in o ma ion in 4 (now 5) languages. Howe e , aiming o
high quali y se ice, Pu elei’s eam wan ed o pe sonalize i s communica ion wi h e e y
cus ome . A he same ime, hey we e awa e ha suppo ing so many eques s wi h adequa e
ca e was labo ious and o en imes in ol ed epe i i e wo k, le alone ime.
Pu elei looked o a way o op imize he execu ion o some o hese asks. Fi s o all, a deep
insigh in o cus ome s’ beha io was needed. Then, wi h enough da a, he suppo eam
planned o au oma e pa o he p ocesses hey we e in ol ed in. Pu elei decided o expand he
help ob ained om Zendesk by in eg a ing i s se ices wi h he ools p o ided by Ul ima e32:
ⓘ
Ul ima e is an AI-powe ed cus ome suppo au oma ion pla o m ha o e s “ he ools,
he eam, and he ac ics ha can ake you om 0 o +60% au oma ion ac oss digi al
suppo channels, including cha , email, messaging, and mo e”. Wi h he help o la ge
language models (LLMs) and gene a i e AI, Ul ima e o e s se ices in up o 109
languages.
The main goal o Pu elei was o c ea e a p oac i e suppo sys em ha would encou age
sel -se ice and ee-up he human agen s, while main aining quali y o pe sonalized cus ome
expe ience.
3.9.3. AI-based solu ion
Ge ing he mos ou o he da a
Zendesk equipped Pu elei's suppo eam wi h a help cen e designed o mul ichannel,
mul ilingual communica ion. Se ice cha , now wo king 24/7, along wi h o he ansac ional
32 h ps://www.ul ima e.ai/
31 h ps://www.zendesk.com/
130 | 249
da a p o ided knowledge abou consume s’ ac i i ies, p e e ences and mos equen ly
encoun e ed p oblems. To access all hese in o ma ion, Zendesk Explo e33 was in oduced:
ⓘ
Zendesk Explo e p ocesses and analyses da a o cus ome expe ience wi hin di e en
communica ion channels o gene a e epo s o eal- ime ac i i ies, encoun e ed
issues and possible in e es s o a pa icula clien . I educes esponse ime, which
esul s in mo e sol ed eques s, and p o ides mo e accu a e con en ha is backed by
insigh s om collec ed da a.
These insigh s became a basis o mos o Pu elei’s cus ome se ice decisions. Ge ing o
know he pa e ns and ends o suppo eques s enabled he employees o adop a p oac i e
a i ude owa ds he clien s. The o eseen ulfillmen challenges o po en ial conce ns could
now be add essed in ad ance, p o iding cus ome s wi h “ he g ea es added alue wi h he
lowes con ac hu dle”.
Au oma ing se ice
Pu elei was now equipped wi h a esponsi e help cen e and a sizable knowledge base. Thei
p oac i e cus ome ca e app oach, howe e , no only equi ed much wo k on he pa o he
suppo eam, bu also in ol ed mos ly epe i i e asks. Despi e ha ing an exhaus i e lis o
FAQs, he help cen e was s ill flooded wi h simple inqui ies.
Al hough Zendesk does include in-house cha bo s in hei o e , i has also a b oad o e o
in eg a ions wi h pa ne companies. Based on he sugges ion by Zendesk’s eam i sel , Pu elei
decided o include Ul ima e’s ools34 in hei wo k:
ⓘ
Ul ima e Zendesk in eg a ion combines ex ensi e CRM knowledge wi h op imizing
capabili ies o i ual agen s. I allows o au oma ed icke handling, in e ac ions wi h
cus ome s and wo kflow moni o ing, while no sac ificing ad anced pe sonalisa ion
coming om da a analysis. Ul ima e's con e sa ional AI is de eloped h ough he
company’s own mul ilingual GPT model.
By choosing he in eg a ion-based solu ion, Pu elei was able o launch hei cha au oma ion
wi hin 3 weeks. Ul ima e, as opposed o simple o ms o au oma ion, was no solely dependen
on p edefined bu ons. In he upg aded cha windows, cus ome s we e able o inpu ee ex ,
as well as images, ha we e hen p ocessed by he bo . This mean ha hey we e able o
engage in much mo e complex con e sa ions.
Due o in eg a ion, Ul ima e had access o Zendesk’s da a, including hei CRM and o de
managemen sys em (OMS). The e o e, cha in e ac ions, al hough au oma ed o some ex en ,
s ill benefi ed om de ailed cus ome in o ma ion o gene a e pe sonalized messages. As o
he simple eques s, hese we e au oma ed comple ely. The cus ome expe ience main ained
i s quali y, all while equi ing less epe i i e e o om he suppo eam.
34 h ps://www.ul ima e.ai/in eg a ions/zendesk
33 h ps://www.zendesk.com/se ice/analy ics/
131 | 249
3.9.4. Resul s
The implemen ed solu ions led o posi i e ou comes on bo h sides o Pu elei’s ansac ions. A
majo i y o clien s op ed o au oma ed sel -se ice, wi h a deflec ion a e su passing 50%.
Resolu ion imes d opped. The employees could now sa e hei a en ion and ime o mo e
complex issues, while cus ome sa is ac ion was claimed o emain s able o e he 93% ma k.
Fu he , ye undisclosed benefi s we e a ibu ed o he accompanying Ul ima e Shopi y
in eg a ion.
3.9.5. Employee skills needed o adop he solu ions
“No de elope s, no coding, no wo ies.” - concludes he p omo ional ideo o Ul ima e one-click
in eg a ion in o Zendesk. Ne e heless, i is wo h no ing ha in he case o Pu elei he
in-house eam included “2 expe s o au oma ion p ocesses and epo ing”. One o hem e en
men ioned he benefi s o he ecognising links be ween he p ocessed cus ome da a, bu did
no speci y i i was done in coope a ion wi h o in spi e o AI ools. As o he p ocess o
implemen ing he solu ion, he eam was “kep in he loop” when i came o he da a impo ed
o es ing.
3.10. Repea
3.10.1. Desc ip ion o he company
Sou ce: h ps://www.sem ush.com/company/s o ies/ epea /
Name
Coun y
Size
Repea
F ance
<50 employees
Desc ip ion: Repea is an e-comme ce company, based in Pa is, selling sus ainable mens ual
unde wea . Mo i a ed by ca e o he en i onmen and cus ome s’ economic s abili y, hey o e
eusable p oduc s o eplace sani a y owels, ampons and cups.
Goal: Launch a new eComme ce b and in paid and o ganic channels.
S a egy: compu e ision, image classifica ion
132 | 249

Technology p o ide : Sem ush
Tools: Backlink Analy ics, PLA Resea ch, Si e Audi , Keywo d O e iew, Keywo d Magic Tool,
SEO W i ing Assis an
3.10.2. The need
Mens ual unde wea in F ance is a ending ma ke . When s a ing ou hei digi al s o e, he
eam behind Repea , wi h 3 membe s and limi ed budge , was acing a challenge o a ac ing
he cus ome s o hei p oduc . They needed a ma ke ing s a egy ha would ensu e a
long- e m each o he company’s o e .
Inexpe ienced in he field o ma ke ing, Repea ounde s ied di e en app oaches and
channels o sell he unde wea . They s a ed ou pa ne ships wi h influence s in paid media bu
wi hou he adequa e insigh in o pe o mances he me hod p o ed o be ime-consuming and
gene a ed oo big o a cos wi hou su icien and imely ROI.
The idea o ma ke h ough paid a ic on Google and Facebook equi ed echnological
knowledge ha he eam was lacking. Wi hou e icien keywo ds, his s a egy would no b ing
he desi ed immedia e esul s.
To go beyond influence s and paid ad e ising, Repea wan ed o channel hei ma ke ing
h ough o ganic a ic. Fo his o be e ec i e, howe e , hey needed a s uc u ed plan ha
would ensu e long- e m sus ainabili y. To analyze he company’s possibili ies and de elop a
s a egy ha would fi hem, he eam eached ou o help om Sem ush35 wi h hei b oad
ange o AI-based ools.
ⓘ
Sem ush is he online isibili y managemen and con en ma ke ing SaaS pla o m. I
p o ides se ices such as websi e audi ing, keywo d esea ch, backlink op imiza ion,
compe i i e analysis, con en c ea ion and anking moni o ing.
The main goal o Repea was o de elop a long-las ing ma ke ing s a egy ha would in ol e
mul iple channels and b ing no able esul s in he websi e's a ic pe o mance.
3.10.3. AI-based solu ion
Finding he e ified influence
To limi he isk o in es ing in he w ong paid channel o communica ion, Sem ush expe s
con inced Repea o sea ch o solu ions ha al eady wo ked o he company’s compe i o s.
35 h ps://www.sem ush.com/
133 | 249
Using Sem ush’s Backlink Analy ics36, he eam was able o sea ch h ough o he fi m’s backlink
po olios and p o ide a lis o si es and blogge s ha e icien ly connec ed use s o e-s o es.
Repea could hen each ou o po en ial pa ne s wi h collabo a ion p oposals. Backlink
analysis also allowed o success ul a ilia e p og ams, in which some pa ne s ecei ed a %
om sales o he p oduc s ha we e pu chased by hei audience h ough he p o ided link.
ⓘ
Backlink Analy ics - composed o a backlink da abase and disco e y ool, i enables
e alua ion and compa ison o link p ofiles o any domain. P o ides up- o-da e insigh
in o backlinks’ ype, li espan, quali y and moni o s ma ke ing s a egies o compe ing
fi ms.
Paid ad e ising
Paid campaigns emained a pa o he s a egy o p o ide ini ial a ic o he websi e.
Al hough he main channel o Repea ’s ad e ising was Facebook, Sem ush ools helped he
company plan and un a success ul campaign in Google Shopping. Sem ush’s PLA Resea ch37
was used o c ea e a lis o aluable keywo ds based on he me ics and s a egies o
compe i o s. The keywo ds could hen be used in copyw i ing o he implemen ed paid
campaigns.
ⓘ
PLA Resea ch - p oduc lis ing ads analysis o op imizing Google Shopping campaigns.
Enables esea ching compe i o s’ ankings, s a egies along wi h de ailed da a on ads
and keywo ds.
O ganic a ic
Compe i ion analysis p o ed ha o he companies we e benefi ing om e icien SEO
s a egies. Domain O e iew38 ga e insigh s in o channels ha gene a ed a ic o Repea ’s
i als. The fi m became con inced ha imp o ing o ganic a ic could uly complemen hei
s a egy and p o ide i s sus ainabili y.
ⓘ
Domain O e iew - p o iding a ull o e iew o a domain and i s online isibili y.
Comp ises da a on ep esen a ion in specific ma ke s, op keywo ds and g ow h ends
o di e en a ic channels. Also includes compe i ion analysis and compa isons by
coun y o ma ke ype.
Building o ganic a ic equi ed se e al efinemen s o Repea ’s websi e. Th ough Si e Audi 39
Sem ush conduc ed checks o si e’s heal h in sea ch o p oblems ha would influence c awling
and Google ankings.
39 h ps://www.sem ush.com/si eaudi /
38 h ps://www.sem ush.com/analy ics/o e iew/?sea chType=domain
37 h ps://www.sem ush.com/analy ics/pla/posi ions/
36 h ps://www.sem ush.com/analy ics/backlinks/
134 | 249
ⓘ
Si e Audi - SEO analysis ool, c awling and checking e e y page on a gi en websi e. I
p oduces egula epo s on websi e’s issues, hei le els o p io i y and solu ions o fix
hem.
Keywo d O e iew and Keywo d Magic Tool40 helped Repea find he keywo ds ma ching hei
ac ual needs. They s a ed wi h niche keywo ds (ma ching sea ches wi h a mo e specific in en )
ha we e easy o be posi ioned high in ankings, hen s a ed a ge ing mo e popula sea ch
e ms.
ⓘ
Keywo d O e iew - keywo d esea ch da abase wi h gene al in o ma ion on a gi en
keywo d, including sea ch olume, di icul y, in en and CPC alue.
ⓘ
Keywo d Magic Tool - opic-based keywo ds sugges ions o op imized and s uc u ed
con en o ganiza ion. Shows possible links be ween mo e gene al opics, subg oups
and ela ed sea ches, including long- ail keywo ds.
Finally, Repea was able o c ea e i s websi e’s con en in line wi h he mos e icien keywo ds
and use s’ com o . Valuable blog pos s could also es ablish Repea as expe s on emale well
being and en i onmen al consciousness. Wi h he help o Sem ush’s SEO W i ing Assis an 41, he
company could gene a e a icles ha we e bo h in e es ing o hei cus ome s and me SEO
equi emen s. The ool p o ided con en b ie s o blog pos s w i e s, checked finished a icles
and e u ned sugges ions o imp o ing hem.
ⓘ
SEO W i ing Assis an - a sma w i ing edi o ha helps op imize copyw i ing o
engagemen and SEO. Assesses he eadabili y o documen s, consis ency o he one
o oice and checks o plagia ism. Sugges ing SEO imp o emen s based on eal- ime
da a.
3.10.4. Resul s
The final s a egy composed o influence ma ke ing, paid a ic and o ganic a ic led Repea
o 3900% a ic g ow h o e 18 mon hs. The o ganic a ic i sel was inc eased by 45% hanks
o jus 10 published blog pos s. Company’s e enue g ew by app ox. 300% and he domes ic
ma ke sha e o Repea became subs an ial.
41 h ps://www.sem ush.com/swa/
40 h ps://www.sem ush.com/analy ics/keywo do e iew/,
h ps://www.sem ush.com/analy ics/keywo dmagic/s a )
135 | 249
3.10.5. Employee skills needed o adop he solu ions
Repea eam membe s we e lacking use ul online ma ke ing knowledge. They s a ed building
hei s a egy in he cou se o ials and e o s. Despi e ecognizing he impo ance o o ganic
a ic, hey we e no able o ge aluable insigh s o pe o m he necessa y imp o emen s o
hei websi e.
3.11. Sinne up
3.11.1. Desc ip ion o he company
Sou ce: h ps:// a le.ai/cus ome s/sinne up
Name
Coun y
Size
Sinne up
Denma k
~250 employees
Desc ip ion: Sinne up is a Danish e aile specializing in in e io design and li es yle p oduc s.
The company’s o e includes popula b ands along wi h hei own in-house designed u ni u e
and home accesso ies. A e o e 50 yea s on he ma ke , Sinne up now uns 14
b ick-and-mo a s o es in Denma k and Ge many and an e-comme ce websi e.
Goal: S able con ol o e cus ome expe ience du ing di e se ma ke ing campaigns.
S a egy: In elligen sea ch unc ions buil in o he websi e.
Technology p o ide s: Ra le
Tools: Ra le AI Sea ch, Ins an Answe s, Ra le Insigh s
3.11.2. The need
Sinne up desc ibes i s cus ome s as he “bigges sou ce o knowledge and inspi a ion” and
openly pu s hei sa is ac ion a he op o p io i ies lis . While i may no be su p ising o hea
such bold s a emen s om a company wi h long adi ions, i ce ainly ele a es expec a ions o
shopping expe ience.
136 | 249
3.13. Velasca
3.13.1. Desc ip ion o he company
Sou ce: h ps://www.domo.com/cus ome s/ elasca
Name
Coun y
Size
Velasca
I aly
<250 employees
Desc ip ion: Velasca is a oo wea and clo hing company based in Milan. The a isanal b and is
de eloped wi h an emphasis on a di ec supply chain, ensu ing quali y, ai alue o wo k and
close ela ionships wi h cus ome s. The p oduc s a e sold in mul iple loca ions ac oss Eu ope,
as well as h ough Velasca’s online s o e.
Goal: Collec ing and o ganizing in o ma ion om di e en sou ces o c ea e da a-d i en
s a egies o u u e de elopmen .
S a egy: Cen alising da a, inc easing i s ope a ional accessibili y and au oma ing key asks.
Technology p o ide : Domo
Tools: Business in elligence and analy ics, Da a in eg a ion, Au oma ion apps
3.13.2. The need
The mission o deli e p oduc s di ec ly om a isans o he clien may seem as a clea p ocess
ha is easily aceable. As Velasca’s ope a ions g ew, howe e , wi h inc easing numbe s o
cus ome s pu chasing shoes and clo hing om di e en loca ions, ge ing an insigh in o he
e iciency o he wo kflow p o ed di icul .
Cus ome da a was indeed collec ed ac oss nume ous online and o line channels, bu lacked
he p ope o ganiza ion ha would p o ide accessible knowledge o he company’s employees.
The widesp ead dis ibu ion o p oduc s was aced wi h di e se demand ends ha had o be
me wi h op imal in en o y managemen and p oduc ion scheduling, le alone accu a e
ma ke ing campaigns.
Making decisions based on he acqui ed da a equi ed ime and human esou ces. Basic
cus ome coho epo s could be done wi hin as much as one week. S ill, he esul s we e no
sa is ying. Employees wo ked wi h incomple e da a and by he ime he analyses we e finished,
143 | 249

hey could ha e al eady been sligh ly ou da ed. Las ly, unin o med supply chain managemen
pu a h ea o he di ec - o-cus ome business model.
Awa e o hese challenges, Velasca seeked a echnological solu ion ha would o ganize he
company’s da a and make i a ailable on ime. This esul ed in he coope a ion wi h Domo Da a
Expe ience Pla o m52.
ⓘ
Domo is “a cloud-based pla o m ha p o ides simplified, nea eal- ime access o
da a.” Deli e ing da a managemen in as uc u e, i enables in eg a ion and con ol
o e in o ma ion. Among i s ools he e a e also low-code and p o-code au oma ion
apps, as well as business in elligence (BI) and analy ics ea u es.
The main goal o Velasca was o in eg a e a da a hub in o i s ope a ions o ad anced,
consis en analy ics and in o med decision making h oughou he o ganiza ion’s uni s.
3.13.3. AI-based solu ion
Accessing in o ma ion
To gain access o all o he company’s da a, Velasca used he help o Domo’s Da a In eg a ion
ea u e53.
ⓘ
Domo’s Da a In eg a ion ools “make dispa a e da a asse s accessible and a ailable o
business analysis”. They enable connec ing la ge olumes o da a om mul iple sou ces
and p ocessing hem wi hin he cloud in as uc u e. Na i e in eg a o s and connec o s
simpli y and speed up accessing he igh in o ma ion a any momen .
Th ough in eg a ion, Velasca was able o b eak down i s da a silos and connec hei con en s o
a single en i onmen . The in o ma ion om e e y online and o line channel could now be easily
synced and sha ed be ween di e en eams. T ansac ional da a, supply chain de ails and
cus ome insigh s we e all cen alized and eady o u he analysis.
Deepe in o da a
Wi h all he da a in he igh place, he company was now capable o ully benefi ing om he
knowledge hey p o ided. Fo he analyses o be done on ime and wi h p ope accu acy,
ano he Domo’s solu ion was in oduced54.
ⓘ
Domo’s Business In elligence and Analy ics enable comp ehensi e da a explo a ion.
Backed by AI algo i hms, hese ools pe o m ad anced analyses o gene a e epo s
54 h ps://www.domo.com/business-in elligence
53 h ps://www.domo.com/da a-in eg a ion
52 h ps://www.domo.com/
144 | 249
and p edic ions. Wi h accessible isualiza ions, he insigh s p o ide a basis o nea
eal- ime decision making.
BI and Analy ics esul ed in mo e accu a e da a ha was a ailable a all imes and almos
ins an ly. Wi h accessible dashboa ds and isualiza ions employees we e able o be e
unde s and he sha ed da a. G oss and ne o de s we e o ganized wi h mul iple me ics, such as
pu chase ime, loca ion and p oduc de ails. In en o y and p oduc ion p ocesses could now be
modified in acco dance wi h any changes in demand.
Cus ome se ice eam gained insigh s abou clien s’ p e e ences, needs and habi s ela ed o
bo h o line and online pu chases. Top-pe o ming p oduc s and majo business d i e s we e
ecognised. Along wi h moni o ing o cu en ma ke ing campaigns, his helped op imize he
company’s o e and ou each o p o ide mo e and be e se ice.
A bi o au oma ion
The las , ye c ucial change o Velasca’s ope a ions was in oduced h ough au oma ion
included in Domo’s App C ea ion Tools55.
ⓘ
Domo’s Business Apps “a e highly cu a ed da a expe iences ha enable anyone o
au oma e business p ocesses and confiden ly ake ac ion using flexible App C ea ion
Tools.” They enable bo h low-code and p o-code app de elopmen using p o ided
amewo ks and in as uc u e. One o he key ea u es is he in eg a ion o au oma ed
wo kflows ha pe o m assigned asks p e iously done manually.
As any o he company, Velasca was o ced o ca y ou epe i i e asks ha engaged esou ces
and ook a significan amoun o ime. Al hough manual p ocessing o da a equi ed cons an
suppo o enginee s i s ill was no ee o human e o s. Au oma ion ook o e pa o he
esponsibili y o connec ing, combining and isualizing da a. Due o ha , ime and wo k o ce
was sa ed o c ea i e and cus ome ca e asks and mo e efined decision making.
3.13. 4. Resul s
The implemen ed solu ions led o he measu able ou comes ha we e accessible hanks o he
Domo’s echnology i sel . A e age i em alue (AIV) became 7.8% highe and a e age o de alue
(AOV) eached 12.3% g ow h. Th ough moni o ing o cus ome li e ime alue, Velasca could
disco e ha in he cou se o 1 yea clien s spen 16.4% mo e on hei o de s. Wi h 23% cos
educ ion and 18% ime sa ed, he company’s o al e enue inc eased by 61% in o e a yea .
55 h ps://www.domo.com/business-apps
145 | 249
3.13.5. Employee skills needed o adop he solu ions
Domo’s ools no only o ganized and p ocessed Velasca’s da a o he company’s p ofi bu also
enabled i s a ailabili y o he wo ke s. Wi h pe sonal dashboa ds and explo a ion ea u es,
employees could unde s and he in o ma ion hey wo ked wi h and imp o e hei da a li e acy.
The need o echnical skills, on he o he hand, appea ed o ha e been educed. Low-code
au oma ion apps enabled pe o ming asks wi h smalle enginee ing eams and less
p og amming ope a ions.
3.14. Bes p ac ices - Inno a i e Re ail Labo a o y
Sou ces: in e iew, in o ma ion collec ed du ing a dedica ed wo kshop o ganized o he INAiR
conso ium by he Inno a i e Re ail Labo a o y, websi e h ps://www.inno a i e- e ail.de/
A special case in he esea ch o AI-powe ed e ail can be ound in Saa b ücken, whe e he
Inno a i e Re ail Labo a o y is loca ed. A pa o he Ge man Resea ch Cen e o A ificial
In elligence (DFKI), he IRL is one o a kind esea ch o ganiza ion, specializing in
"applica ion-o ien ed de elopmen s o he e ail o he u u e”. As a public-p i a e pa ne ship,
hey coope a e wi h go e nmen o ganiza ions and uni e si ies, while main aining a s ong
connec ion o he business. While he IRL conduc s he majo i y o ope a ions in he esea ch
cen e s in Saa b ücken and Be lin he employees also a el h ough he coun y, p esen ing
and explaining newly de eloped solu ions.
Since 2007 The Labo a o y has been con ibu ing o he implemen a ion o new echnologies
and me hods in he e ail indus y, wi h a ificial in elligence as an impo an poin o ocus. I s
in e disciplina y eam is composed o expe s in enginee ing, p og amming and psychology,
ensu ing ha any inno a ion would mee a complex se o demands o he ma ke , socie y and
go e nmen . The IRL is e y keen o use AI as a esea ch ool. Wi h me hods based on ad anced
analy ics, machine lea ning, deep lea ning, compu e ision and cha bo s, he o ganiza ion aims
o esol e complex p oblems o he e ail indus y.
The mul idisciplina y eam goes h ough he en i e p ocess o de eloping a new cus om
echnology. They conduc echnology scou ing in sea ch o opics ending ac oss e ail
businesses. Fi s ideas can be ma e ialized in he o m o p o o ypes o MVPs, which a e hen
es ed and con on ed wi h he po en ial use s. The mix o heo e ical and p ac ical knowledge
146 | 249
esul ing om he wo k o he IRL employees is no only used in e nally bu sha ed u he ,
du ing he wo kshops and aining sessions conduc ed by expe s.
While he scope o he Inno a i e Resea ch Labo a o y is b oad, any pa o he desc ibed AI
de elopmen can become a poin o ocus. An especially in e es ing elemen o he Lab’s o e
a e i s Resea ch as a Se ice (RaaS) p ac ices. Companies ha lack knowledge o esou ces o
es new ideas can each ou o he Labo a o y’s help. I hey ask a specific ques ion, IRL is
he e o find an answe .
The wo k o pa icula impo ance o INAIR is he Mi els and-Digi al Zen um Handel - a
sec ion o he Ge man go e nmen ’s ini ia i e o digi alising small and medium-sized
companies, in his case ocusing on he e ail sec o . The Inno a i e Re ail Labo a o y is
esponsible o he AI- ela ed pa s o he esea ch, including wo kshops, p esen a ions and an
educa ional podcas . Wi h financial suppo om Mi els and-Digi al, he IRL eam also
conduc s implemen a ion p ojec s o e ail SMEs.
The Inno a i e Re ail Lab ocuses on esol ing social and en i onmen al challenges in e ail.
They de elop sys ems ha encou age consume s o shop wi h hei own packaging and moni o
hei home in en o y wi h sma s o age shel es. O he expe imen s aim o edefine e ail
s o es as “social mee ing poin s,” making hem mo e inclusi e o elde ly people. Many
inno a ions a e designed o amilia ize consume s wi h AI echnologies h ough un and
engaging expe iences, such as VR and AR en i onmen s o gamified shopping expe iences.
Wi h p ac ical p ojec s like finge -poin - igge ed p oduc ecogni ion and in elligen ui
c a es, he Inno a i e Re ail Lab aims o s ay a he o e on o digi al inno a ions in he e ail
sec o . Howe e , he o ganiza ion no es ha many companies fi s need simple solu ions.
Reluc ance owa ds echnological change o en means businesses ope a e wi hou s uc u ed
en e p ise esou ce planning (ERP) o da a awa eness. Nume ous imp o emen s a e needed
be o e a ificial in elligence can become a p ima y ocus.
The ollowing opics a e he p ima y ocus o he IRL:
● Analysis o consume beha io
● Au oma ed in en o y main enance
● Da a p o ec ion and p i acy
● Influence o he In e ne o Things on socie y
147 | 249
● Inno a i e p edic ion sys ems
● Inno a i e ma ke ing o use -o ien ed se ices and dialogue ma ke ing
● In elligen assis ance sys ems
● Machine Lea ning and senso usion
● Na iga ion and kiosk sys ems in he supe ma ke
● Pe sonalized and mobile pu chasing suppo
● Robo ics and logis ics suppo
● Sma labels (e.g. RFID and NFC) in e ail
● Con ol o cus ome flows
● Technologies o enhance he cus ome expe ience in depa men s o es
● Fusion o digi al and analog wo lds
Sou ce: Inno a i e Re ail Labo a o y p esen a ion.
Figu e 37 O e iew o he Inno a i e Re ail Lab’s ac i i ies
Sou ce: Inno a i e Re ail Labo a o y p esen a ion.
148 | 249

3.15. Bes p ac ices - Foo p in s AI & Danubius
Sou ces: h ps://www.mis .com/wp-con en /uploads/Foo p in s-AI-In o.pd ,
h ps://www. omanianbusinessjou nal. o/ oo p in s-ai- omanias- e ail-media-ma ke - o-exce
ed-10-million-eu os-in-2024/,
h ps://danubius.o g/en/danubius-si- oo p in s-ai-lanseaza- e eaua-de- e ail-media-cu-cea-
mai-ma e-acope i e-geog afica-din- omania/
Ano he in e es ing p ojec comes om Romania and is no limi ed o one o ganiza ion, bu is
he esul o a ui ul collabo a ion be ween wo local echnology companies. One o hem
de elops ad anced so wa e, while he o he is ocused on supplying e aile s wi h ha dwa e.
Toge he , hey managed o include Romanian SMEs in he new en i onmen o B2C business,
benefi ing om inno a ions in a ificial in elligence.
Foo p in s AI is an AI-based omnichannel e ail media pla o m specializing in dynamic
a ge ing based on consume beha io . Th ough da a-d i en au oma ion, p edic i e beha io al
models and hype local ac ionable insigh s, he company enables analysis o anonymous a ic,
le e aging he po en ial o da a om digi al and physical poin s o con ac alike. I enables
b icks-and-mo a e aile s o op imize hei ma ke ing s a egies in a way ha was p e iously
only possible in e-comme ce.
Figu e 38 Foo p in s AI’s cus ome da a gene a ion pa e n
Sou ce: h ps://www.mis .com/wp-con en /uploads/Foo p in s-AI-In o.pd .
149 | 249
Danubius p esen s i sel as “one o he mos impo an in eg a o s o ax and paymen solu ions
in Romania”. Du ing i s 30 yea s in he ma ke , he company has been o e ing elec onic cash
egis e s ha in oduced he mos ecen inno a ions in fiscal paymen s (ca d ansac ions,
blue oo h, OS in eg a ions, po able p in e s, BlueCash). I does no come as a su p ise ha a
some poin , i would c oss i s pa h wi h he de elopmen o AI.
The solu ion b ough abou by Foo p in s p o ed success ul o he pla o m’s in e na ional
pa ne s. The clien s, howe e , we e mos ly la ge e aile s wi h adequa e esou ces. Meanwhile,
he b oad o e o Danubius allowed he ECR p o ide o each di e en kinds o businesses,
which esul ed in a sizable ne wo k o me chan s, including he smalle ones. The cash egis e s
u ned ou o be he key o in oducing AI echnology o Romanian SMEs.
The e ec o he collabo a ion was a new ype o sma cash egis e . A ailable as a ee
upg ade o all he clien s possessing he newes model o Danubius’ p oduc , his equipmen is
able o ga he ansac ional da a and connec i o Foo p in s’ sys em. The AI algo i hms enable
p ofiling o he cus ome s based on hei ac i i ies and gene a e p edic ions on u u e beha io
and p e e ences. The esul ing ma ke ing s a egies can be deployed wi hin he e ail media
ne wo k, launched oge he by Foo p in s and Danubius as an al e na i e o popula ad e ising
pla o ms (Google, Me a) and adi ional media ( adio, TV). Impo an ly, all he cus ome da a is
said o be co ec ly anonymised, in line wi h he GDPR policies.
As one can see om he scheme abo e, cash egis e s a e only one o many po en ial sou ces o
cus ome da a, sugges ing ha his solu ion does no ye ealize he ull po en ial o Foo p in s
AI’s sys em. I is, ne e heless, an impo an s ep owa ds democ a iza ion o he a ificial
in elligence echnologies, allowing SMEs wi h physical s o es o en e he compe i ion ueled by
he digi al ans o ma ion o e ail. Danubius CEO s a ed ha he e aile s pa icipa ing in he
fi s s age o p oduc de elopmen can expec o “benefi om an addi ional income o 25%,
subjec o ce ain condi ions”.
3.16. Bes p ac ices - Da e Media
Sou ces: h ps://www.da emedia.pl/, in e iew wi h Da e Media’s CEO
Polish e-ma ke ing company is an example o a di e en app oach owa ds AI adop ion which
may be in e es ing o e ail businesses jus se ing o on hei ad en u e wi h a ificial
in elligence echnologies. In sho , Da e Media specializes in “g ow h hacking s a egies”
h ough da a analysis and au oma ion, based on ex ensi e use o popula so wa e. Tools
p o ided by Google Analy ics, OpenAI, Sem ush o Sales o ce a e jus a ew o he as ange o
solu ions ha he fi m has o o e . I s a eas o in e es include cus ome suppo ,
150 | 249
communica ion channels, SEO, ma ke ing campaigns and da a-d i en op imiza ion.
I may seem coun e in ui i e o add a middleman on op o all he financial cons ain s ha
s and in he way o MSMEs’ digi al ans o ma ion. Da e Media’s s eng h, howe e , lies in i s
expe ise - he knowledge o whe e o s a and how o do i . The company’s se ice begins wi h
an audi and a se o ecommenda ions on possible imp o emen s. To minimize he isk o
was ed esou ces, many collabo a ions ope a e on a success- ee model, whe e he echnology
p o ide ea ns a sha e o he p ofi s ha esul om imp o emen s. Ini ial es ing and
scalabili y also help smalle businesses ha a e no ye con inced owa ds AI solu ions.
As a ma ke ing- ocused company, Da e Media uses a ange o ools o con en gene a ion.
Specifically, ans o ming one kind o con en in o ano he is a p ocess ha can be easily
accele a ed and simplified h ough au oma ion. Da e Media’s CEO men ions Cha GPT
(h ps://openai.com/cha gp /), Claude (h ps://claude.ai/) and Gemini
(h ps://gemini.google.com/) as e y handy when i comes o edis ibu ing media. These ools
enable gene a ing ansc ip s o audio files, social media pos s and blog a icles.
O he solu ions employed by Da e Media a e audio isual gene a ion so wa e o p e-p oduc ion
(s o yboa ds, anima ics), edi ing (highligh s, e ec s) and e en c ea ing whole films o songs
om sc a ch. In hese cases, he fi m ecommends ools such as Luma
(h ps://lumalabs.ai/d eam-machine), Midjou ney (h ps://www.midjou ney.com/home),
Runway (h ps:// unwayml.com/), S o yboa de (h ps://s o yboa de .ai/), Suno AI
(h ps://suno.com/) and Topaz (h ps://www. opazlabs.com/).
In some ins ances Da e Media uses i s subsc ip ions o he abo emen ioned se ices o
gene a e a b oad, comp ehensi e o e o i s own clien s. Whole sys ems o au oma ed CRM,
SEO, sales o paid ad e isemen may be chosen, o , as is o en he case wi h MSMEs, smalle
imp o emen s emain a lowe -budge op ion. Da e Media claims o be keeping ack o he
la es inno a ions and upda ing hei oolse s acco dingly wi h new bu e ified so wa e
a ailable on he ma ke .
When i comes o da a analysis, howe e , he e is mo e inpu om he company i sel . Based on
Cha GPT echnology, Da e Media eam c ea es hei own sc ip s ha enable collec ing and
p ocessing cus ome da a om di e en channels. Re iews sp ead all o e he social media,
o ums and e-s o es can be easily o ganized and pu oge he wi h ma ke analysis, ueling
in o med decision making as well as sales o ma ke ing s a egies.
151 | 249
Acco ding o Da e Media’s CEO, e en he mo e echnologically ad anced fi ms may need
“guidance on in eg a ing AI and digi al ools in o hei ope a ions”. Toge he wi h au oma ion
ea s and conce ns abou da a secu i y, hese skill gaps should be add essed h ough “ aining
and eassu ance”. Ge ing a p o essional o e iew and es ing o possible AI implemen a ion
may be a good fi s s ep o MSMEs jus en e ing he new e a o e ail. A e ha , i is o hem o
decide whe he o keep ou sou cing he whole se ice o con inue he upskilling p ocess o
de eloping in-house solu ions.
3.17. Towa ds classifica ion o models o AI adop ion and
in eg a ion in e ail companies
Table 10 Main echnologies and applica ion a eas by use case
Company
name
AI echnology applied
AI applica ions in he
alue chain
Benefi s o AI applica ion
ALOHAS
• Cha bo
• Augmen ed
in elligence
• Cus ome engagemen
• Ope a ions
op imisa ion
• Good se ice (g ea e a ailabili y and
sho e esponse ime due o cha bo
assis ance)
• Au oma ion o cus ome se ice
(answe ing equen inqui ies)
Feed.
• Cha bo
• Insigh engine
• Speech ecogni ion
• Augmen ed
in elligence
• Cus ome engagemen
• In en o y managemen
• Knowledge and insigh
managemen
• Ope a ions
op imisa ion
• Good se ice (pe sonalisa ion,
omnichannel connec ion)
• Au oma ion o cus ome se ice
(o de upda es, esol ing simple
issues)
• Op imisa ion (wo kflow moni o ing,
chain supply moni o ing wi h
eal- ime epo s)
G ünewald
• ML
• Deep lea ning
• RPA
• In en o y managemen
• Knowledge and insigh
managemen
• Ope a ions
op imisa ion
• P edic ion ( o ecas ing demand and
sales ends)
• Au oma ion (gene a ing o de s
ma ching demand)
Ke igans
• Cha bo
• Augmen ed
in elligence
• Cus ome engagemen
• Knowledge and insigh
managemen
• Au oma ion ( ecommenda ions,
changes in s ock and p ices)
• Good se ice (pe sonalisa ion,
152 | 249