IBMInstituteforBusinessValue|ResearchInsights

Cybersecurity2028

Yourworkforce,builtfortheAIfrontier

aws

Contents

Beatingthetickingclock

3

Stageone:Crawl

7

Stagetwo:Walk

17

Stagethree:Run

22

Thepowerofstartingwhereyouare

27

AbouttheAWS+IBMpartnership

IBMConsultingbringsbusinessandITtransformationexperiencetogetherwithdeepindustryexpertise,

methods,frameworks,andgenerativeAI-powered

assetsandassistantstoaccelerateyourhybridcloud

andAIjourneyontheAWSCloud.IBM’sexpertisein

security,enterprisescalability,andopeninnovation

withecosystempartnershelpyourbusinesstooperatewithspeed,scale,andtrust.

AsanAWSPremierTierpartner,IBMConsulting

CybersecurityServiceshelpssecurecomplex

workloadsrunningonAWSandacceleratehybridcloudstrategybyusingcriticalsecuritycapabilitiesfrombothIBMandAWS.

Ourpartnershipempowersbusinessestoinnovate,

optimize,andscalewithconfidenceinarapidlyevolvingdigitallandscape.Bycombiningtrustedexpertisewithleadingsoftwareandservices,IBMandAWSenable

organizationstoseizeunprecedentedopportunitiestoachievetransformative,business-drivenoutcomesatscale.

Formoreinformation,pleasevisit:

/aws/security

1

“Oneofthebiggestthreats

isnotanyadversary,butourownlackofimagination.”

KoosLodewijkx,VicePresident,CISO,IBM

Key

takeaways

Security’snewcoreisAI.

AIisnowastrategiccapabilityatthecenterofthe

operatingmodel,transforminghowtechandsecurity

teamsworktogether.Overthenextthreeyears,AI

augmentationisexpectedtoincrease50%whiletheuseofgenAIsecuritycapabilitiesisexpectedtogrow63%Nearlytwo-thirdsofexecutives(64%)expectevery

employeeintheirIT/ISorganizationtobeusingAIagentswithinthenexttwoyears.

●A36-monthsprintwillseparatetheAIresilientfromtheAIdisrupted.

Organizationsareinoneofthreestatesatthemomentvis-à-visAI-centricsecurityoperations:crawl,walk,

orrun.18%arecrawling,52%arewalking,and30%oforganizationsarerunningonthefasttrack.

AI-firsttransformationcanpayforitself.

Notonlyintermsofbudgetbutalsointermsofeffort.

Executivesestimatetheycansaveanaverageof10-20%oftheirtotalcybersecuritybudgetthroughtheadoptionofadvancedAIusecases.Seepage25formoredetails.

2

Alookatourstudy

7

Averageannualrevenueofsurveyedorganizations

sectors

andindustries

~$21B

17

countriesacrossAsiaPacific,

Europe,MiddleEast,LatinAmerica,andNorthAmerica

Over

CISOs,CTOs,andCIOs

550

COOs,andCHROs

450

CEOs,CFOs,

Chiefinformationsecurityofficers(CISOs)havespearheadedthe

evolutionofdigitalsecurity,convertingeverytechnologicalupheaval

intoanopportunitytostrengthentheirorganizations’defenses.

Theplaybooksformostsecurityandtechnologyleaderswerewrittenforthegrowthopportunitiesofcloudmigrationsanddigitalservices—temperedbytheexpandingattacksurfacesthesecreated.Andyet,

manyoftheassumptionsthatunderlietheseplaybooksarebecomingobsolete.Why?

First,human-intermediatedoperationsarebeingreplacedbyAI-intermediated

technologies.1Andsecond,frontierAImodelsareevolvingsorapidlyleaderscan’tforeseewithanyconfidencewherecapabilitieswillbeintheneartomid-term—

complicatingtheirplanningandinvestmentdecisions.2

Whatisclear:advancedAImodalitiesarechanginghoworganizationsworkin

fundamentalways—creatingopportunitiesbutalsointroducinguncertaintiesand

risks.3Forinstance,morethanoneinfourAIinitiativeshavebeencancelled,

postponed,orfailedtoscalebecauseofsecurityconcerns,andmorethanoneinthreeorganizationsindicatetheirAIcapabilitieshavealreadybeencompromisedby

cyberattacks.4TheuncheckedgrowthofnonhumanidentitiesfueledbygenerativeAIactuallyexpandstheattacksurface,createsnewtargetsforthreatactors,andleadstoalossofvisibilityintowhoorwhatisaccessingcriticalsystems.5

Fittingly,AItechnologycanhelpCISOswithAIthreats.However,insteadoftraditionaloperationsenhancedwithAI,AIismovingtothecenteroftheoperatingmodel—

evolvingintoastrategiccapabilitythatisredefininghowtechnologyandsecurity

teamsworktogetherandcollapsingthedistancebetweenbusinessoperationsand

businessoutcomes.Yetourresearchrevealssecurityandoperationsleadersarelivinginahousedivided,withsentimentsplitonwhethertheirorganizationalcultureis

inhibiting(44%)ratherthanenablingchange(56%).

Oneconclusionfromouranalysis:thenext36monthsrepresentacriticalwindowofopportunityforCISOs.WebelievehowleadersrespondduringthistimewillseparatetheAI-resilientfromtheAI-disrupted.

Beatingthetickingclock:The36-monthsprint

Inthisreport,wemapoutthetransformationjourney

towardAI-firstoperations.Hint:cybersecurityworkforcedevelopmentandchangestotheIT/ISoperatingmodel

willplayabigrole.Organizationsareinoneofthreestatesatthemoment:crawl,walk,orrun.

Beatingthetickingclock

Stageone:Crawl

Stagetwo:Walk

Stagethree:Run

Thepowerofstarting

whereyouare

Roughlyoneinfiveenterprisesare

in“crawl”mode

Basedoninsightsfromover1,000security,technology,andbusinessexecutives

across17geographiesandsevenindustrysectors(seeresearchmethodologyon

page29),aminorityoforganizationsarestillintheearlystagesofmodernizingtheir

cybersecuritydeliveryandsupportmodel.Ouranalysisrevealsapproximatelyoneinfiveorganizations(18%)arestillina“crawl”statewhenitcomestoAItransformation.Theseenterprisesarelargelyfocusedonplanningandincrementalchange,withlittletoshowfortheirtransformationeffortstodate.

Themajorityarein“walk”mode

Thefutureisall“run”

Mostorganizations(52%)arefindingtheirway,typicallyfollowingoneoftwo

paths:prioritizingAI-firstworkforcetransformationorenhancingIT/ISintegration.

Wecharacterizetheseorganizationsasbeinginthe“walk”phaseoftheir

transformationjourney,whereleadersareactivatingtheirworkforceandbuilding

momentumforchange.ThedifferencefortheseorganizationsisthatAIispositionedtobecometheconnectivetissuerunningthroughouttheirentireorganization.

30%oforganizationshavemovedthefastesttowardbuildingtheirAI-firstfoundation.Theseorganizationsareenteringthe“run”phaseoftheirtransformationjourney,

wherecybersecuritycapabilitiesachieveanewlevelofvisibility,reach,andautonomy.Inthisstate,IT/ISoperationshaveevolvedtobecomemoreself-regulating,

self-correcting,andself-healing,withtheabilitytodelivernotonlyautomated

(scripted)responsesbuteffectivelyself-initiatedandgoal-orientedcybersecurity

outcomes.Intherunstate,cybersecuritycapabilitiesarelargelyautonomous—

formulatingdecisions,testingtheirownhypotheses,andleveragingagenticAItobothenrichdecisioningandautomateinference-drivenactions.Theseorganizationsare

adaptingrapidlytochangesintheoperationsenvironment,openingthedoortogreaterresilience,innovation,andgrowth—safelyandatscale.

“Werecentlyhadaconversationwithoneofthe

leadersofthenewfrontierAIfirms.Hesaidthat

overthepasttwoyears,thecapabilitiesofmodelshavedoubledevery10months.Ifthatgrowth

curvecontinues,howlongwillittakeforthesecapabilitiestobe10timeswhattheyaretoday?It’saboutthreeyears.Thatputsusin2028.”

KoosLodewijkx,VicePresident,CISO,IBM

3

Perspective

Cybersecurity’sAIboomischangingalltherules

Thisisn’tjustanothertechtremor;theAIrevolutionisaseismicevent,

afundamentalrewritingoftherulesofdigitalengagement.6Thispresentsbothopportunitiesandrisks.Accordingtoonestudy,thecomplexityof

tasks“generalistAI”cancompletehasbeendoublingroughlyeverysevenmonthsforthelastsixyears.7

ExecutivesanticipatenewAImodalitieswillgrowsignificantlyoverthenextthreeyears:AIaugmentationisexpectedtoincrease50%,theuseofgenAIsecuritycapabilitiesisexpectedtogrow63%,withAIagents(48%)and

workflowautomationandorchestration(45%)expectedtoplayamoreprominentroleinday-to-dayoperations.

Buttofullyseizetheseopportunities,CISOsmustmanagethreedistinct

paradigmshifts—andallatthesametime.Tacklingthesedemands

simultaneously—notsequentially—isessentialtosecuringtheorganizationfor2028andbeyond.

AI:Newuses,newgrowth

AIagents

(autonomousAI)

AIaugmentation(usingAIsolutionstoenhanceorextendhumanskills)

+50%48%

+48%35%

32%

24%

TodayIn3years

TodayIn3years

GenAI

capabilities

Workflowautomation&

orchestrationcapabilities

+45%42%

+63%28%

29%

17%

TodayIn3years

TodayIn3years

Qs:Estimatethepercentageofcybersecurityworkloadsthatrelyuponthefollowingcapabilities—Today(n=1013).

Estimatethepercentageofcybersecurityworkloadsthatwillrelyuponthefollowingcapabilities—In3years(n=1013).Note:Itemsarenotmutuallyexclusivesopercentagesdonotsumto100%

4

Beatingthetickingclock

Perspective(continued)

Cybersecurity’sAIboomischangingalltherules

First,organizationsareracingtogetaheadof

AI-enabledthreatactorswhoareapplyingnew

capabilitiestolearn,adapt,andevolvewith

relentlessspeedandsophistication.8Forgetsimplysecuringwheredatalivesorhowapplicationsare

secured;AIisrewritingwhoorganizationsare

defendingagainstandwhatmakesthesplit-seconddecisionsthatdeterminethetrustworthinessand

resilienceoftheirdigitaloperations.Indeed,

cybercriminalsarealreadyleveragingAItools

tooutmaneuverandoutpaceenterprisedefenses,

usingadaptivemalware,deepfakes,andautomatedattackstocreateandexploitnewvulnerabilities.

Second,CISOsmustaddressthedemandfornewAIservicesacrosstheenterprise.9AsAIbecomestheconnectivetissueineveryprocessandfunction,howtheyenablesecureandtrustedAIneedsto

change.ThenextwaveofAIinnovation—multiagentsystems,largeactionmodels,andsyntheticdata—promisestoredefinewhat’spossible,pushingAI

deeperintomission-criticaloperations.

Andthird,CISOsaregrapplingwithhowtouseAIfortheirownoperations.10Fromautomation’sefficiency,togenerativeAI’screativeproblem-

solving,toagenticAI’sautonomousactions,AIisrapidlybecomingthecybersecurityteam’smostpowerfulally.

WhileAIcapabilitiesaresurging,manyenterprisesarestalledbyungovernedsprawlandtangled

clouds.11Thecorequestionforsecurityandtechleaders:cantheypivotfastenoughtobuildAI

operationsatscale,orwilllegacyinfrastructureandoperationsderailtheirAItransformation?

“Afewweeksago,Ihadameetingwithabunchofstart-ups.Iaskedhowmanyofthemweretrainingtheirownlarge

languagemodels.Therewereafewdoingsoinfocused

areas.Butwhatsurprisedmewashowmanyofthemwere

not.Theytoldmelargelanguagemodelsareevolvingsofastthatwhentheypausetodotheirownfinetuningortraining,theyfindthattheircapabilityfallsbehindthegeneral

purposemodels.Theevolutionishappeningamazinglyfast.Idon’tthinkthatthisgrowthwillcontinueforeverbutthat’sthephasewe’reinnow.We’rejustseeingmassivechangesmonthovermonth.”

ChrisBetz,ChiefInformationSecurityOfficer,AWS

Beatingthetickingclock

5

Perspective

Alitmustestforcyberresiliencein2028

Threequestionstoaskyourself.

Overthenextthreeyears,cybersecurityoperationswillreorientaroundnewAIcapabilitiesthatspantheIT/ISportfolio.Belowarethreeareaswherechangeislikelytobesignificant.

“Weknowthatthingsaregoing

tobeverydifferentin2028.Are

businessesgoingtotakeadvantageofthis?I’mgoingtogiveyouan

answerthatIexpecttobeentirelyright:someareandsomearen’t.”

ChrisBetz,ChiefInformationSecurityOfficer,AWS

DoesyourAIprovidetotalvisibility?

IT/ISobservabilityandresiliencegobeyondtraditionalmonitoring.Theyareabout

leveragingAItoprovideadeep,real-timeunderstandingofyourentireIT/IS

environment.Bycontinuouslyanalyzingvaststreamsoflogs,metrics,andtracesfrom

everysystem,AIcandeliverunprecedentedvisibilityintosystembehaviorand

performance.Thisenablesproactiveidentificationofanomalies,predictive

troubleshooting,andtheabilitytooptimizesystemsformaximumavailabilityand

reliability.It’sthefoundationfortrulyunderstandingandstrengtheningyourdigital

posture.ObservabilityenablesITandISteamstoproactivelymanageandoptimizetheirsystems,helpingensurehighavailability,reliability,andperformance.

DoesyourAIdeliversmarteroperations?

AIOpsisn’tjustaboutautomation;it’saboutinfusingintelligenceintoyourITand

securityoperations.ItinvolvesusingAIandmachinelearningtoanalyzemassivedatasetsfromdiverseITsystems—logs,metrics,events,andmore—toautomatically

detectcomplexanomalies,predictpotentialissuesbeforetheyimpactoperations,

andautomatecriticalincidentresponses.Thissignificantlyimprovesoperational

efficiency,reducesmeantimetoresolution(MTTR),andfreesyourteamstofocusonstrategicinitiativesratherthanreactivefirefighting.

CanyourAIself-direct

forsuccessfuloutcomes?

Autonomouscybersecuritysolutionsrepresenttheevolutiontoself-managing

security—usingAI,machinelearning,andadvancedautomationtodetect,respondto,andmitigatecyberthreatswithminimalornohumanintervention.Beyondsimple

threatblocking,thesesystemscanautonomouslyidentifyvulnerabilities,analyze

threatsinrealtime,andtakedecisiveactions—suchascontaininganintrusionor

enforcingpolicycontrols.Thisdramaticallyenhancesthespeed,efficiency,and

scalabilityofyoursecurityoperations,allowingforcontinuousdefenseagainstrapidlyevolvingthreats.

6Beatingthetickingclock

7

Stageone:Crawl

AIoverload:

Findingsignalinthenoise

ForalltheinterestandinvestmentinAI,thereisanequalamountofconfusionandindecision.Securityconcernshaveemerged

Beatingthetickingclock

Stageone:Crawl

Stagetwo:Walk

Stagethree:Run

Thepowerofstarting

whereyouare

asaprimaryobstacletowidescaleenterpriseAIadoption.12

Onehurdleisaperceptionthatinnovationandsecurityaresomehowatodds—

creatingsomestrategicparalysisamongtechnologyexecutiveswhomaybethinkingtheyneedtochoosebetweenthetwo.13Opinionisevenlydividedonwhethernew

AIgovernanceframeworkscreateexcessivefriction,with49%sayingtheydo.

Similarly,leadersaresplitonwhetherorganizationalcultureisimpeding

transformation,with44%seeingtheircultureasabarrierratherthananenabler.

Furtherskepticismisevidentfromthe62%ofexecutiveswhosaytheirorganizationsareplacingexcessivefaithinAI’stransformativepotential.Thisfindingunderscoresagrowingimpatienceamongstakeholders:it’stimefortheoreticalbenefitsto

translateintomeasurablebusinessoutcomes.

Lookingahead,autonomousAIagentsrepresentthemostpromisingdevelopmentonthenear-termhorizon:

76%

72%

67%

64%

ofexecutivesanticipateAIagentswillfundamentallyimproveoperationswithintwoyears.

viewAIagentsaskeycatalystsfororganizationalinnovation.

anticipateagentswillsignificantlyenhanceROIonexistingAIinvestments.

expectuniversaladoptionofAIagentsamongIT/ISstaffinthesametimeframe.

AgenticAI:

Thenextfrontier

TheseinsightssuggestthatwhiletheAIsecuritylandscaperemainschallenging,

executivesareidentifyingspecifictechnologies—particularlyautonomousagents—thatcanjustifyacceleratinginvestmentinAIinfrastructuredespitethehurdles.Inother

words,someIT/IS,business,andoperationsleadersarefindingsignalinthenoise.

FIGURE1

LeadersexpressoptimismaboutAIagents

“AIagentswill

signi?cantlyimprove

76%

thewayweoperate”

72%

“AIagentswill

unlockinnovation

inourorganization”

“AIagentswillsigni?cantlyimprovethereturnwegetonourAIinvestments”

67%

64%

“Weexpecteveryemployeeinourorganizationtobe

usingAIagentsin2years”

“ExecutivesinourorganizationhavesetclearguidelinesandmeasurableobjectivesforAIassistantsandagents”

61%

Q:Towhatextentdoyouagree

withfollowingstatementsabout

theuseofAIagentswithinyourIT/ISorganization?(n=1013)

“Weneedeverybodytobethinking

aboutthisandbeverydeeply

involvedinwhatAIiscapableofandwhereit’sgoing.Whereweuseit

today,wherewecan’tuseittoday,wherewecanuseitinsixmonths

whenawholenewsetofusecasesbecomeavailabletous.BecauseI

thinkthecapabilitiesareadvancingfasterthanpeople’simagination.”

KoosLodewijkx,VicePresident,CISO,IBM

8Stageone:Crawl

BridgingtheAIknowledgegap

ManyorganizationsareinacrawlstatenotonlybecauseofissuesaroundgovernanceandROI,butalsobecausetheyneedtopreparetheirsecurityworkforce.Somearemakingsignificantstridesinworkforcepreparation—butatroublingdisparitybetweenexecutiveandemployeereadiness

threatenslong-termsuccess.

ThetalentequationhasemergedasanequallycriticalfactorinAI

implementation.AccordingtoarecentIBMIBVstudyofC-levelexecutives,talent,skills,andcybersecurityaretheleadingchallengesorganizationswillfaceoverthenextthreeyears.14

FIGURE2

Executivespointtotalent,

60%

skills,andcybersecurity

astheirbiggestchallenges

Talentrecruitingandretention

Leadingchallenges

57%

overthenext3years

Skills

54%

Cybersecurity

challengesoverthenextthreeyears?Chooseupto6.(“2025Skillsand

o:haiilietpresent

talent”dataset,IBMInstituteforBusinessValue,n=2690)

Stageone:Crawl9

FIGURE3

Top-heavyAI:

Readinessskewed

towardexecutives,

notemployees

Organizationshavemovedswiftlytoaddresssecurityworkforceneeds,with65%establishingformal

ExecutivesEmployees

(non-managerial)

upskillingplanstosupporttheirAIstrategy.Nearly

asmany(63%)havelaunchedstructuredchange

managementprogramstointegrateAIassistants

andagents,while60%havedeliveredformaltrainingonhowAIwillreshapedailywork.

82%

Yetbeneaththeseencouragingmetricslies

aconcerningreality:AIreadinessremains

disproportionatelyconcentratedattheexecutivelevel.Thedatarevealsasignificantdisparity—

82%ofexecutivesdemonstratefluencyinAI

capabilities,limitations,andresponsibleuse

practices,comparedtojust53%offrontline

53%

employees.Similarly,while52%oforganizationshaveembeddedAIskillsintoexecutive

developmentpathways,only38%haveextendedtheseopportunitiestotheirbroaderworkforce.

Thisasymmetryisn’tjustatalentissue—itrepresentsanexistentialthreattoAItransformationitself.

Whenthepeopleactuallyusingthesetoolslack

fundamentalunderstanding,eventhemost

Understand

thecapabilities,

limitations,and

responsibleuseofAIassistantsandagents

sophisticatedAIimplementationswillinevitably

falterinpractice.Closingtheexecutive-employeeAIfluencygapoverthenext36monthsisanessentialbrickinthefoundationuponwhichsustainableAI

transformationmustbebuilt.

“Weknowthattherearegoingtoberightwaysandwrongways

andwedon’tknowwhatallthosehiddentrapsareyet.”

ChrisBetz,ChiefInformationSecurityOfficer,AWS

38%

IntegrateAIskillsinto

professionaldevelopmentplans/pathsfor

theseroles

52%

10Stageone:Crawl

Perspective

AdoptingsecurityABCsspeedsoperatingmodeltransformation

Securityawareness,behaviors,andculturepractices(securityABCs)aretheunderpinningforAI-centricoperatingmodeltransformation.ManyCISOsstillhaveworktodointhisarea.Forthemostpart,onlyaminorityofCISOsdescribetheirorganizationsashavingrobust

capabilitiesinAI-relatedsecurityABCs.

Awareness

Speci?crisksassociated

withAI(suchasprompt

manipulation,manipulatedcontent)

46%

OrganizationAI-relatedpoliciesandguidelines

43%

PotentialforAI-relatedprivacyandintellectualpropertyviolations

40%

Ethicalconsiderations

aroundfairnessandbias

37%

AIregulationsand

guidelines(suchasEUIsAIAct)

33%

Behaviors

Protectingsensitivedata

andpreventingdataleakage

50%

RestrictingaccesstoAI

systemsbasedonprincipleofleastprivilege

50%

MonitoringandloggingofAIsystemstodetectanomaliesandunusualbehavior

41%

Conductingredteam

activities,suchaspenetrationtestingandsimulatedattacks

38%

Securecodingpractices

forAImodeldevelopment

36%

Culture

Promotingashared

responsibilitymindset

withusers,stakeholders,andpartners

48%

Integratingsecurity

capabilitiesintoAIassetsandservicesacrosstheorg

43%

AdoptingAIethicsstandards

43%

LeadershipadvocacyofAI-relatedinitiatives

41%

Mandatingeducation

andawarenessprogramsforAI-relatedrisks

36%

Q:ConsideringyourIT/ISorganization’sabilitytocounterAI-relatedthreats,assesyourcurrentcapability

acrossthefollowingfactors:[Percentagesrepresentresponsesfor“Wide-scalecapability(mostusecases)”and“Comprehensivecapability(virtuallyallusecases)”](n=1013).Note:Toassesstherelativesecurity

maturityoftheorganization,theabovefactorswerecompiledintoa“SecurityABCs”index.Indexscoreswerecombinedwithotherfactorstoassesstheorganization’sAItransformationprogress,asdescribedin“Researchmethodology”onpage29.

Stageone:Crawl11

12Stageone:Crawl

Perspective

AdoptingsecurityABCsspeedsoperatingmodel

transformation

Feedingthe

securityflywheel

Securityawareness,

behaviors,andculturereinforceeachother,buildingmomentum

forchange.

Culture

Behaviors

Awareness

Outcomes

“You’vegottokeepfeedingtheflywheel,butonceyougetitgoing,itkeepsmomentumthatwillhelpyoudealwiththebumpsyouencounteralongtheway.IthinkofsecurityABCsastheflywheelthathelpspowerthingsforward,thathelpsusadapt.”

ChrisBetz,ChiefInformationSecurityOfficer,AWS

Stageone:Crawl13

Casestudy

Singapore’sstrategyforpublicsectorgenAIadoption

Singapore’sGovernmentTechnologyAgency(GovTech),responsibleforthenation’sdigitalpublicservices,hascreatedamodelforenterprise-scalegenerativeAI

adoption.15GovTechsawgenerativeAI’spotentialtotransformpublicservices,fromstreamliningprocessestopersonalizingcitizeninteractions.Amajorchallengewasthehighcostofrunninglargelanguagemodels(LLMs)atanationallevel.

GovTech’ssolution(akaMAESTRO)isdesignedtodelivercost-efficient,pre-built,andproduction-readygenerativeAIcapabilitiesacrossgovernmentagencies.

JeffreyChai,MAESTROProductManageratGovTech,notedthestrategicimportance:“MakinggenerativeAImoreaccessibleandsustainableforouragenciesiscritical.

Thisallowsustouseitspotentialwhilemanagingresourcesresponsibly.”

GovTech’splatformandadoptionmodelofferablueprintforCISOsandexecutivesglobally.Theirexperienceshowsthatfocusingonoptimizedcoststructuresand

accessibleplatformsfacilitatesscalingofgenerativeAI,deliveringsignificantoperationalimprovementsandbettercitizenservicesefficiently.

Keyoutcomesandstrategicapproaches

–75%improvedcostperformanceforgenAIworkloads.GovTechoptimizeditscomputinginfrastructureforgenerativeAI’sdemands.

Theyselectedhigh-performingfoundationmodelsandusedtechniquessuchasmodelquantization.AmazonBedrockandSageMakerJumpStart

allowedthemtoselectright-sizedmodels,and

thendeploymultiplesmaller,specializedmodelsinsteadoflarge,generalones.Thisapproach

significantlycutresourceconsumptionandcostswithoutlosingcapability.

–Acceleratedusecasesandbroadaccessibility.

Theplatformfeaturesano-code,unified,web-

basedinterfacethatsimplifiesMLmodelbuilding,training,anddeployment.Thismadecomplex

machinelearningaccessibletonontechnicalstaff.Byremovingtechnicalbarriers,GovTechrapidly

increasedgenerativeAIuseacrossthe

government.Withinninemonths,theplatformwasadoptedby20publicsectororganizations,involvingover45projectteamsandmorethan

300datascientistsandMLengineers.

Real-worldimpactand

operationaltransformation

GovTech’splatforminvestmentisdeliveringclearbenefits:

–MinistryofManpower(MOM).Usedtheplatform

tobuildanAI“sensemaker”tool.Itprocessedoveronemilliondocumentsinthreemonths,boosted

insightsextractionby60%,andcutsensemakingtimeby50%,savingover2,000workhours.MOMalsodeployedanautomatedjobclassificationtoolthatprocessed10millionjobpostingswith92%

accuracyinthreemonths.

–CentralProvidentFundBoard(CPFB).Usedtheplatformtosummarizetranscriptsfromabout

600,000annualcitizencalls.TheseAI-generatedsummarieshelpwithfollow-upsandidentifying

emergingpublicissues,improvingservicequalityandoperations.

14Stageone:Crawl

Actionguide

Securityinmotion

Breakoutbycommitting

toanAI-firstsecuritystrategy

AI-firsttransformationleaders(CISOs,CTOs,CIOs)

Commit.ConvertAIexperimentationintoAIinnovationandexecution.Understandthe

prioritiesofyourbusinessandoperationscounterpartsanddiscusshowdistributingtechnologyandsecurityexpertisecould

improveefficiency.

UseAIasacatalyst.CreateaholisticviewofAI

capabilitiesacrossinfrastructureandoperations.“Solveonce”byarchitectingAI,cloud,and

securityinconcertandatscale,thenmodernizeyourtechnologyandsecurityoperations

accordingtoAI-firstprinciples.

Connecteverything.Buildbridgesacrossthe

organizationbyconnectingsecurityABCstogoalsandoutcomes.Workonbuildingmomentumto

makefasterdecisionsandtopowerchange.

Becomeabridgebuilder.Focusoncreating

abettervaluepropositionbyconnectingcore

securityprinciplessuchastrust,integrity,and

resiliencewithbusinessobjectivessuchas

innovation,speed,andgrowth.Rallyaroundthesetocreateaf