2376 - How GenAI + watsonx Code Assistant can help with in-place modernization _1728936986121001lySD.PDF

1、How GenAI+watsonx code assistant for z can help in-place mainframe appmodernizationhype vs.value Session 2376Oct 2024IBM TechXchangeDevOpsDevOps never was about tools,neither is AI.Tools only enable itEveryone thinks they know everything,and no one knows a damn thing.Elizabeth StroutPeople who think

2、 they know everything are a great annoyance to those of us who do.Isaac AsimovPeople who think they know everything really annoy those of us who know we dont.Bjarne StroustrupOne additional phenomenon with AI is that due to its availability to the wide public,almost everyone has some experience with

3、 it.Set Expectations-Create a common understandingWhat LLM is the right oneIs there a special LLM for the job at hand?Can you use the latest LLM?Define scope(in and out-of scope)Agree where AI will be used(and where not)Where does AI provide a benefitWhat are associated risksPrepare for expected out

4、comes,fail-fast is also an acceptable result!Source:https:/ Gartner Hype Cycle Its in phase where its real benefits are being evaluated AI offerings on Mainframes can be categorized broadly into three categoriesAI Offering on MainframeCore AI OfferingsGenAI OfferingsMachine Learning on ZWatsonx Code

5、 Assistant for z and LLM modelsWatsonx Assistant for zAI is the assistant,the enabler if you want.Its not the driver.Human intelligence remains essential for success.Client Legacy Code,ArtifactsAccenture Industry Data/Process DomainInputsEmbedModelEmbedModelMappingPromptDesign/TemplateLLMsMemory Sto

6、reResponseScoringBusiness Strategy Roadmap:OutputsOutputsFunctional,Technical Documentation,Test CasesKnowledge Graph of Legacy Code baseDomain TaggingReverse EngineerReverse EngineerLegendData/Process FlowKey InputIterative8Business Strategy RoadmapWatson Code Assistant for zAccentures AI powered r