2024代理型AI：生成式人工智能(GenAI)的新前沿 -Agentic AI – the new frontier in GenAI

AgenticAI

–thenewfrontierin

GenAI

Anexecutiveplaybook

HarnessingAIisn’tjustabout

technology—it’saboutunleashingunprecedentedpotential.

Inanerawherespeed,e?ciency,andcustomercentricitydictatemarketleadership,organisationsneedto

harnesseverytoolattheirdisposal.Overthepastcoupleofyears,arti?cialintelligence(AI)hasexplodedontotheworldstage,withcompaniesandindividualsacrossthegloberapidlyadoptingthetechnology.TheGCCisplayingaleadroleinthespace,withbusinessleadersintheregionexploringwaysofintegratingthisrapidly

developingtechnologyintotheiroperations.

GenerativeAI(GenAI)isbeingrecognisedasagame-changerforinnovationintheregion,empoweringenterprisesbyautomatingroutinetasks,enhancingcustomerexperiencesandassistingincritical

decision-makingprocesses.Insightsfromour27thAnnualCEOSurvey:MiddleEast?ndingshaveshownthat73%ofCEOsintheMiddleEastbelieveGenAIwillsigni?cantlychangethewaytheircompanycreates,deliversandcapturesvalueoverthenextthreeyears

1

.GenAIispoisedtomakeasigni?canteconomicimpact,with

estimatesindicatingthatitcouldcontributebetween$2.6trillionand$4.4trillionannuallytoglobalGDPacrossvariousindustriesby2030.Inspeci?csectors,suchasenergy,investmentsinGenAIareexpectedtotriple,

from$40billionin2023toover$140billionbytheendofthedecade.Thissurgeininvestmentre?ectsthe

transformativepotentialofGenAI,particularlyinenhancingproductivity,streamliningbusinessprocesses,andreshapingvaluechainsacrossindustries

2

.

Againstthisbackdrop,multimodalGenAIagenticframeworkshasemergedastransformativecatalysts,

enablingbusinessestoaccelerateprocessautomationatanunprecedentedscale.ThistechnologyinvolvesmultipleAIagentsworkingtogether,eachspecialisingindi?erenttasksordatatypes,tosolvecomplex

problemsandautomateprocesses.Bycollaboratingandconstantlylearning,theseagentsenhance

decision-making,optimiseprocesses,anddriveinnovation.ItcombinesrangeofadvancedAItechniquestoprocessdiversedatatypesandautomatecomplextasks.

Thecentralquestionisn’twhethertoadoptthistechnology,buthowswiftlyorganisationscanintegrateittostayaheadofthecompetition.Thisexecutiveplaybookexploreshoworganisationscanleveragethis

technologytoboostoperationale?ciency,enhancecustomerexperience,anddriverevenuegrowth.Itprovidesreal-worldsuccessstoriesspanningindustrysectorsandorganisationalfunctions,strategicinsights,tactical

blueprints,andbestpracticestoguideyourjourneyintothisrevolutionarylandscape.

Keyinsights

●AgenticAI,di?erentiatedbyitsadvancedhuman-likereasoningandinteractioncapabilities,is

transformingthemanufacturing,healthcare,?nance,retail,transportation,andenergysectors,amongothers.

●Organisations’AIstrategiesshouldleveragemultimodalGenAIcapabilitieswhileensuringethicalAI

safeguardstodriveautonomousprocessre-engineeringandenhanceddecision-makingacrossalllinesofbusiness.

●Integratede?ectively,agenticAIcanenhancee?ciency,lowercosts,improvecustomerexperience,and

driverevenuegrowth.

WhatisagenticAI?

AgenticAIgenerallyreferstoAIsystemsthatpossessthecapacitytomakeautonomousdecisionsandtakeactionstoachievespeci?c

goalswithlimitedornodirecthumanintervention

3

.

KeyaspectsofagenticAI

Goal-orientedbehaviour:TheseAIagentsaredesignedtopursuespeci?cobjectives,optimising

theiractionstoachievethedesiredoutcomes.

Autonomy:AgenticAIsystemscanoperateindependently,

makingdecisionsbasedontheirprogramming,learning,and

environmentalinputs.

Environmentinteraction:An

agenticAIinteractswithits

surroundings,perceivingchangesandadaptingitsstrategies

accordingly.

Work?owoptimisation:AgenticAIagentsenhancework?owsandbusinessprocessesbyintegratinglanguageunderstandingwith

reasoning,planning,and

decision-making.Thisinvolvesoptimisingresourceallocation,improvingcommunicationandcollaboration,andidentifyingautomationopportunities.

Learningcapability:ManyagenticAIsystemsemploymachinelearningor

reinforcementlearningtechniquestoimprovetheirperformanceovertime.

Multi-agentandsystem

conversation:AgenticAI

facilitatescommunication

betweendi?erentagentsto

constructcomplexwork?ows.Itcanalsointegratewithother

systemsortools,suchasemail,codeexecutors,orsearch

engines,toperformavarietyoftasks.

Learningcapability

Environmentinteraction

Work?ow

optimisation

Goal-oriented

behaviour

Autonomy

Multi-agentandsystemconversation

EvolutiontomultimodalGenAIagents

InAI,theonlyconstantischange—embraceacultureofperpetualinnovation.

Thejourneyofagenticframeworksbeganassimple,rule-basedsystemsdesignedtoperformspeci?ctasks.Overtime,thesesystemshaveevolvedintosophisticated,multimodalagentscapableofprocessingandintegratinginformationfromvarioussources,suchastext,images,andaudio.MultimodalitycapabilitiesallowAIagentstounderstand,employ

reasoning,andinteractlikehumans,enhancingtheire?ectivenessandversatilitytosolveawiderangeofbusinessproblems

4

.

Theevolutioncanbebrokendownintothreekeyphases:

(2000s)

IntegrationofMachineLearning(ML)

○Learningfromdata:TheintegrationofMLallowedagentstolearnfromlargedatasets,improvingtheirabilitytomakedecisionsandperformtasks.Thiswasasigni?cantstepforwardfromrule-basedsystems,asagentscouldnowadapttonewinformationandimproveovertime.

○NaturalLanguageProcessing(NLP)enableduserinteractions:AdvancesinNLPenabledagentstounderstandandgeneratehumanlanguagemoree?ectively,makinginteractionsmorenaturalandintuitive.

(2010s)

Introductionofmultimodality

○Combiningtext,images,andaudio:Multimodalagentsemerged,capableofprocessingandintegrating

informationfromvarioussources.Forinstance,anagentcouldanalyseatextdescription,recogniseobjectsinanimage,andunderstandspokencommands.Thismultimodalitymadeagentsmoreversatileandcapableofhandlingcomplextasks.

○Enhanceduserinteractions:Multimodalagentscouldinteractwithusersinmoredynamicways,suchasprovidingvisualaidsinresponsetotextqueriesorunderstandingcontextfromacombinationofspokenandvisualinputs.

2020s-present

Advancedautonomyandreal-timeinteractions

○Advancedautonomy:Agentscanoperateindependently,rationaliseandsettheirowngoals,developpath(s)toattainthesegoals,andmakeindependentdecisionswithoutconstanthumanintervention,leveragingdatafrommultiplesourcesorsyntheticdatasets.Inamulti-agenticorchestrationsystem,the?rstsetofagents

focusonmimickinghumanbehaviour(e.g.ChatGPT-4o),thatis,thinkingfasttocomeupwithsolution

approach,whilethesecondsetofagentsfocusonslowreasoning(e.g.ChatGPT-1o)tocomeupwithavettedsolution

5

.Combiningthinkingfastandslowreasoning,agentscanprocessinformationandmakeoptimal

decisionsinreal-time–crucialforapplicationslikeautonomousvehicles,real-timecustomerservice,and

variousmission-criticalbusinessprocesses.ThisautonomymakesagenticAIparticularlypowerfulindynamicandcomplexreal-worldenvironments.

○UserinteractionswithinanethicalandresponsibleAI-controlledenvironment:Withincreased

capabilities,therehasalsobeenafocusonensuringthatagenticsystemsoperateethicallyandresponsibly,consideringfactorssuchasbias,transparency,andaccountability.

IntegrationofML(2000s)

Learningfromdata

NLPenableduserinteractions

AIagent

Goal-orientedbehaviour

Introductionofmultimodality(2010s)

Combiningtext,images,andaudio

Enhanceduserinteractions

Advancedautonomyandreal-timeinteractions(2020s-present)

UserinteractionswithinanethicalandresponsibleAI-controlledenvironment

Human-likereasoningandadvancedautonomy

Whyorganisationsshouldpayattention

Inthefastlaneoftechnologicalevolution,missingtheAIturntodaymeansbeingoutpacedtomorrow.

AgenticAIo?erssigni?cantadvantagesine?ciency,decision-making,andcustomerinteraction.Byautomatingroutinetasksandprovidingintelligentinsights,agenticAIcanhelporganisationssavetime,reducecost,andimproveoverall

productivity.Moreover,organisationswhoadoptanagenticAIsystemcangainacompetitiveadvantagebyleveragingitscapabilitiestoinnovateandenhancetheirbusinessoperations.Lowercosttoentryandeconomiesofscalemakesit

favourablefororganisationstofullyharnessthecapabilitiesito?erscomparedtoitspredecessorsliketraditionalMLandRoboticProcessAutomation(RPA)-drivenautomations.

AgenticAIsystemscansigni?cantlyenhanceanorganisation’scompetitiveedgebyautomatingcomplexwork?ows,

reducingoperationalcosts,andimprovingdecision-makingprocesses.Thesesystemsaredesignedtoadapttochangingbusinessenvironments,drivinghigherproductivityandenablingorganisationstostaycompetitive.Forexample,agenticAIcanpredictmarkettrendsandcustomerpreferences,allowingbusinessestotailortheirstrategiesproactively.This

adaptabilitynotonlyimprovese?ciencybutalsofostersinnovation,givingcompaniesasigni?cantedgeovercompetitors

6

.

Moreover,agenticAIsystemscanhandlelargevolumesofdataandextractactionableinsights,whichcanbeusedtooptimiseoperationsandenhancecustomerexperiences.Byautomatingroutinetasks,thesesystemsfreeuphumanresourcestofocusonmorestrategicinitiatives,therebyincreasingoverallorganisationalagilityandresponsiveness

7

.

Enhanceddecision-making

AgenticAIsystemscananalysevastamountsofdataquicklyandaccurately,providingvaluableinsightstoinformbetterdecision-making.Businessescanleveragetheseinsightstooptimiserevenueandoperations,identifymarkettrends,andmakedata-drivendecisions.Forinstance,inthe?nancialsector,AIcananalysemarketdatatopredicttrends,inform

investmentstrategies,andboostinvestmentROI.Inretail,itcanstreamlineinventorymanagementbypredictingdemandandoptimisingstocklevels.

Boostede?ciencyandproductivity

AgenticAIcansigni?cantlyenhancebusinesse?ciencyandproductivitybyautomatingroutinetasksandprocesses.Thisallowsemployeestofocusonmorestrategicandcreativeactivities.Forexample,incustomerservice,agenticAIcan

handlecommoninquiries,freeinguphumanagentstotacklemorecomplexissues.Inmanufacturing,AI-drivenrobotscanmanagerepetitivetaskswithprecisionandconsistency,reducingerrorsandincreasingoutput.

Improvedcustomerexperience

ByintegratingagenticAI,businessescano?erpersonalisedandresponsivecustomerexperiences.AI-drivenchatbotsandvirtualassistantscanprovideinstantsupport,answerqueries,andevenrecommendproductsbasedoncustomer

preferencesanddynamicinteractions.Thisimprovescustomersatisfaction,buildsloyalty,anddrivessales.Forexample,e-commerceplatformsuseAItorecommendproductsbasedonbrowsinghistoryandpurchasebehaviour.

HowtoconceptualiseagenticAI

solutionsforfuturebusinessoperations

AgenticAIsystemsarerede?ningcustomerservicecentresandaregainingpopularityasagame-changingcapabilityforbothgovernmententitiesandprivatesectororganisations.Whiletraditionalrule-basedchatbots

(software-as-a-service)providedbasic24/7support,andRetrievalAugmentedGenerated(RAG)-basedchatbots

enhancedhuman-likeinteractions(enhancedsoftware-as-a-service),agenticAIsurpassesbothintermsofaccuracy,contextualcoherence,andproblem-solvingability.

Intermsofaccuracy,rule-basedchatbotsarelimitedtoprogrammedresponses,causinginaccuracieswhenqueries

falloutsideofprede?nedrules.RAG-basedchatbotsdependonretrieveddatathatmaynotmatchuserintent.In

contrast,thenovelapproachofagenticAIallowsittounderstandnuancesinlanguage,generatingaccurateresponseseventocomplexorunseenqueries.Itsabilitytolearnfromvastdatasetsenhancesprecisionandadaptability,makingitsuperiorforcustomerinteractions.

Oneofthebiggestlimitationsofchatbotshasbeencontextualcoherence.Rule-basedchatbotsstruggletomaintain

contextinextendedinteractionsduetolinearscripting,leadingtodisjointedresponsesthatharmcustomer

experience.RAG-basedchatbotsmayproduceinconsistentrepliesifretrievalmechanismsdon'tconsiderpreviousinteractions.WhereasagenticAI’sorchestrationcapabilityhelpsitexcelattrackingconversationhistory,

understandingdialogue?ow,ensuringresponsesremaincontextuallyappropriateandcoherent,signi?cantlyboostingcustomerengagement.

Thusfar,bothrule-basedandRAG-basedchatbotshavelimitedautonomousproblem-solvingability.Theformercan'thandleproblemsoutsidetheirscriptswhilethelatterprovideinformationbutcan'tsynthesisedataandpreparethe

human-liveproblem-solvinglogictosolvecomplexissuesacrossintegratedsourcessuchasCRMs,ERP,orIVR

systems.TheagenticAIperformsdynamicreasoninganddecision-making,leveragingaseriesofautonomousagents,analysingcustomerissues,consideringmultiplefactors,andapplyinglearnedknowledgetoresolveproblemsmore

e?ciently.Theoutcomeisquicker,solution-oriented,and?uidconversationsthatenhancecustomerexperienceandsetnewstandardsfore?ciencyandresponsivenessinautomatedcustomerservice.

Micro-agentsOrchestratoragentMasteragent

Customersupportagent

Customersupportagent

User

experience

agent

Issue

resolution

agent

Feedback

collection

agent

FAQagent

Nthagent

Statusupdatesagent

AgenticAIbusinessimperatives

Organisationsmanagingday-to-dayoperationsstandtogainsigni?cantlyfromagenticAIsystems,embracingthe

emerging"service-as-a-software"model.Thisinnovativeapproachtransformsmanuallabourintoautomated,AI-drivenservices.Ratherthanpurchasingtraditionalsoftwarelicencesorsubscribingtocloud-basedsoftware-as-a-service

(SaaS),businessescannowpayforspeci?coutcomesdeliveredbyAIagents.Forexample,acompanymightemployAIcustomersupportagentslikeSierratoresolveissuesontheirwebsites,payingperresolutionratherthanmaintainingacostlyhumansupportteam.Thismodelallowsorganisationstoaccessawiderrangeofservices–whetherit’slegalsupportfromAI-poweredlawyers,continuouscybersecuritytestingbyAIpenetrationtesters,orautomatedCRM

management–atafractionofthecost.Thisnotonlydrivese?ciencybutalsosigni?cantlyreducesoperationaloverheads.

Byleveragingtheservice-as-a-softwaremodel,businessescanautomatebothroutineandhighlyspecialisedtasksthatwereoncetime-consuming,requiredskilledprofessionals,andtypicallyinvolvedexpensivesoftwarelicencesorcloud

solutions.AIapplicationswithadvancedreasoningcapabilitiescannowhandlecomplextasks,fromsoftware

engineeringtorunningcustomercarecentres,enablingcompaniestoscaletheiroperationswithoutaproportionalincreaseincost.Thistransitionexpandstheservicesavailabletoorganisationsofallsizes,freeingthemtofocusonstrategicprioritieswhileAIsystemsmanagetheoperationalburden.AdoptingtheseAI-drivenservicespositions

businessestostaycompetitiveinanever-evolvingmarketplace

8

.

Transitioningfromcopilottoautopilotmodels

Service-as-a-softwarerepresentsanoutcome-focused,strategicshift,enablingorganisationstotransitionfromtheircurrentstatetooperatingin"copilot"andultimately"autopilot"modes.Sierra,forinstance,o?ersasafetynetby

escalatingcomplexcustomerissuestohumanagentswhennecessary,ensuringaseamlesscustomerexperience.WhilenotallAIsolutionso?erthisbuilt-infallback,acommonstrategyistoinitiallydeployAIina"copilot"role

alongsidehumanworkers.Thishuman-in-the-loopapproachhelpsorganisationsbuildtrustinAIcapabilitiesovertime.AsAIsystemsdemonstratetheirreliability,businessescancon?dentlytransitiontoan"autopilot"mode,whereAI

operatesautonomously,enhancinge?ciencyandreducingtheneedforhumanoversight.GitHubCopilotisaprimeexampleofthis,assistingdevelopersandpotentiallyautomatingmoretasksasitevolves.

OutsourcingworkthroughAIservices

Fororganisationswithhighoperationalcosts,outsourcingspeci?ctaskstoAIservicesthatguaranteeconcrete

outcomesisanincreasinglyattractiveoption.TakeSierra,forexample:businessesintegrateSierraintotheircustomersupportsystemstoe?cientlymanagecustomerqueries.Insteadofpayingforsoftwarelicencesorcloud-based

services,theypaySierrabasedonthenumberofsuccessfulresolutions.Thisoutcome-basedmodelalignscostsdirectlywiththeresultsdelivered,allowingorganisationstoharnessAIforspeci?ctasksandpaysolelyforthe

outcomesachieved.

ThisshiftfromtraditionalsoftwarelicencesorcloudSaaStoservice-as-a-softwareistransformativeinseveralways:

Targetingservicepro?ts:TraditionalSaaSfocusedonsellinguserseats,whereasservice-as-a-softwaretapsintoservicepro?tpools,deliveringsolutionsthatfocusonspeci?cbusinessoutcomes.

Outcome-basedpricing:Insteadofchargingperuserorseat,service-as-a-softwareadoptsapricingmodelbasedontheactualoutcomesachieved,directlyaligningcostswithresults.

High-touchdeliverymodels:Service-as-a-softwareo?ersatop-down,highlypersonalisedapproach,providingtrusted,tailoredsolutionsthatmeetthespeci?coperationalneedsofbusinesses.

Whyshouldorganisationsconsiderearlyadoptionandavoidbeinglatemovers?

Latemovers

Earlyadopters

Struggletocatchupandmissoutoncreatingcompetitiveadvantage.

SlowtoinnovatebusinessprocessesandtakefulladvantageofAIsolutionstocreatedi?erentiation.

Playcatch-uptomatchthepersonalisedservicesofearlyadopters.

Higherlostopportunitycostduetolateentryandadoptions.

Missoutonearlylearningopportunitiesandindustryin?uence.

Struggletoachievesimilarmarketshare.

Facehigherbarrierstoentryduetoestablishedcompetitors.

Payrelativelylowercostofentryandlowerlearningandexperiments.

Marketposition

Innovation

Customer

relationships

Operationale?ciency

Learningcurve

Marketshare

Barrierstoentry

Costtoentry

Setindustrybenchmarks

andgain?rst-movermarketadvantage.

LeverageAItoinnovatebusiness

processes,deploytheAIsolutionse?ectivelyandcreatedi?erentiation.

Builddeepercustomerrelationshipsthroughpersonalisedandnewer

experiences.

Streamlineoperationsandreduceoperationalcostearlyon.

Bene?tfromtheinitiallearningcurveandshapeindustrystandards.

Increasemarketshareandpro?tabilitythroughearlyadoption.

CreatebarriersforcompetitorsthroughdeepAIintegration.

Payrelativelyhighercostofentryanditerativetest-and-learnduetonewAIsolutions.

Real-worldsuccessstories

Catalysingchangeacrossallindustries

Manufacturing:SiemensAG

SiemenstransformeditsmaintenanceoperationsbydeployingAImodelsthatanalysesensordatafrommachinery.Thesystempredictsequipmentfailuresbeforetheyoccur,schedulingmaintenanceproactively.Themultimodalframeworkprocessesdatafromvarioussources–vibration,temperature,andacousticsignals–providingaholisticviewof

equipmenthealthandproactivemaintenanceorchestratedbytheagenticAImodels.

Financialimpact:

●Savings:Reducedmaintenancecostsby20%

●Revenuegrowth:Increasedproductionuptimeby15%

Non-?nancialbene?ts:

●Enhancedequipmentreliability

●Improvedworkersafety

Technologystack:

●AImodels:Regressionanddeeplearningmodels

●Platforms:SiemensMindSphere

9

●Tools:Scikit-learn,TensorFlow,Keras,IoTsensors

Healthcare:MayoClinic

ByintegratingAIintoitsradiologywork?ows,MayoClinicallowsforquickerandmoreaccuratediagnoses.ThemultimodalAIprocessesimagingdataalongsidepatienthistoryandlabresults,o?eringcomprehensiveinsightsthataidradiologistsindecision-making,automatingdocumentationandprocessautomationacrosstheradiologyvaluechain.

Financialimpact:

●E?ciencygains:Reduceddiagnostictimesby30%

●Costreduction:Lowered

unnecessaryproceduresby15%

Non-?nancialbene?ts:

●Improveddiagnosticaccuracy

●Enhancedpatientoutcomes

Technologystack:

●AIModels:Regressionand

ConvolutionalNeuralNetworks(CNNs)models

●Frameworks:NVIDIAClaraplatform

10

●Tools:Scikit-learn,PyTorch,MedicalImagingData

Finance:JPMorganChase

JPMorgan’sContractIntelligence(COiN)platformusesAItoanalyselegaldocuments,extractingkeydatapointsin

seconds.Themultimodalframeworkinterpretscomplexlegallanguage,images,andtables,streamliningaprocessthatoncetookthousandsofhumanhours.

Financialimpact:

●Savings:Saved360,000hoursofmanualreviewannually

●Riskmitigation:Signi?cantlyreducedcompliancerisk

Non-?nancialbene?ts:

●Enhancedaccuracyindocumentanalysis

●Improvedemployeeproductivity

Technologystack:

●AImodels:NLPwithGenerativePre-trainedTransformers(GPT)

●Frameworks:COiNplatform

11

●Tools:Python,Hadoop

Retail:Amazon

AmazonleveragesAItoanalysebrowsingbehaviour,purchasehistory,andevenvisualpreferences.MultimodalAImodelsgeneratepersonalisedrecommendations,orchestratetasksacrossorderful?lmentvaluechains,andenhancethe

shoppingexperiencetodrivesales.

Financialimpact:

●Revenueboost:Increasedsalesby35%throughpersonalised

recommendationsandone-clickorderful?lment

●Customerretention:Improvedloyaltyratesby20%

Non-?nancialbene?ts:

●Enhancedcustomersatisfaction

●Increasedengagementtimeontheplatform

Technologystack:

●AImodels:RegressionanddeeplearningModels

●Frameworks:Amazon

Personalise

12

andAmazonOrderFul?lment

●Tools:AWSSageMaker

Transportationandlogistics:DHL

DHLutilisesAImodelstopredictandorchestrateshippingdemands,optimiseroutes,andmanagewarehouseoperations.Thesystemprocessesdatafromvarioussources,includingtra?cpatterns,weatherconditions,andordervolumes.

Financialimpact:

●Costsavings:Reducedoperationalcostsby15%

●E?ciencygains:Improveddeliverytimesby20%

Non-?nancialbene?ts:

●Enhancedcustomersatisfaction

●Reducedcarbonfootprint

Technologystack:

●AImodels:MLmodelsandrouteoptimisationalgorithms

●Frameworks:DHLResilientsupplychainplatform

13

●Tools:IoTdevices,MLmodels

Energy:BP(BritishPetroleum)

BPusesAItoanalyseseismicdata,generating3Dmodelsofsubterraneanstructures.Themultimodalapproachcombinesgeological,geophysical,andhistoricaldatatoidentifyfavourabledrillingsitesandorchestratedrillingequipmentsettings

foroptimaloutcomes.

Financialimpact:

●Savings:Reducedexplorationcostsby20%

●Revenuegrowth:Increased

successfuldrillingoperationsby15%

Non-?nancialbene?ts:

●Reducedenvironmentalimpact

●Improvedsafetymeasures

Technologystack:

●AImodels:RegressionandGenAImodels

●Frameworks:Azurecloudservices

14

●Tools:MicrosoftAI

Education:Pearson

Pearson’sAImodelstailoreducationalcontenttoindividuallearnerneeds,adjustingdi?cultylevelsandcontenttypesbasedonperformanceandengagementdata.

Financialimpact:

●Revenueincrease:Boostedcourseenrollmentby25%

●Costreduction:Lowered

contentdevelopmentcostsby15%

Non-?nancialbene?ts:

●Improvedstudentoutcomes

●Enhanceduserengagement

Technologystack:

●AImodels:Adaptivelearningalgorithms

●Frameworks:Multimodalcontentdeliverysystems

15

●Tools:Python,TensorFlow

Mediaandentertainment:Net?ix

Net?ixusesAImodelstorecommendandorchestratecontentbyanalysingviewinghabits,ratings,andevenvisual

contentfeatures.Themulti-modalAIensuresthatusers?ndcontentthatresonateswiththeirpreferences,keepingthemengaged.

Financialimpact:

●Subscribergrowth:Increasedretentionratesby10%

●Revenueboost:Enhanced

engagementleadingtohighersubscriptionrenewals

Non-?nancialbene?ts:

●Personaliseduserexperiences

●Improvedcontentstrategy

Technologystack:

●AImodels:MLandGenAImodels

●Frameworks:Net?ixmultimodaluserinteractionanalysis

16

●Tools:AWS,ApacheSpark

Telecommunications:AT&T

AT&T’sAImodelsanalyseandorchestratenetworkperformancedataandcustomerinteractionstooptimisenetworkoperationsandpersonalisecustomerservicethroughchatbots.

Financialimpact:

●Costsavings:Reduced

operationalexpensesby15%

●Revenuegrowth:Improved

upsellingthroughpersonalisedo?ers

Non-?nancialbene?ts:

●Enhancednetworkreliability

●Improvedcustomersatisfaction

Technologystack:

●AImodels:MLfornetworkanalytics

●Frameworks:Edgecomputingwithmultimodaldatainputs

17

●Tools:AIchatbots,dataanalyticsplatforms

Governmentandpublicsector:SingaporeGovernment

SingaporeutilisesAImodelstoorchestrateandmanagetra?c?ow,energyconsumption,andpublicsafety.The

multi-modalsystemprocessesdatafromvarioussensorsandcitizenfeedbackmechanismstomakereal-timedecisions.

Financialimpact:

●E?ciencygains:Reducedadministrativecostsby25%

●Economicgrowth:AttractedUS$12billioninforeign

investment

Non-?nancialbene?ts:

●Improvedpublicservices

●Enhancedqualityoflifeforcitizens

Technologystack:

●AImodels:MLandGenAImodels

●Frameworks:SmartNationplatform

18

●Tools:IoTsensors,cloudcomputing

Real-worldsuccessstories

Innovationwithinbusinessfunctions

Humanresources:Unilever

UnileverusesAItoscreencandidatesbyanalysingvideointerviewsandresponses,allowingrecruiterstofocusonthemostpromisingapplicants.

Financialimpact:

●Costreduction:SavedoverUS$1millionannuallyin

recruitmentcosts

●E?ciencygains:Reducedhiringtimeby75%

Non-?nancialbene?ts:

●Enhanceddiversityinhiring

●Improvedcandidateexperience

Technologystack:

●AImodels:NLPandfacialrecognitionalgorithms

●Frameworks:Multimodalcandidateassessmentplatforms

19

●Tools:HireVueAIplatform

Customerservice:BankofAmerica

Erica,anAIvirtualagent,handlesoveramillioncustomerqueriesdaily–includingsnapshotsofmonth-to-datespendingand?aggingrecurringc