2847 - 智能体人工智能和 IBM i.pdf

1、Orlando,FLOctober 69IBM TechXchange 2025Explore.Build.Launch.Jesse GorzinskiSenior Business Architect,IBM Ashish ThomasSoftware Development Engineer,PC RichardSession 2847Agentic AI and IBM iAI that can AI that can generate for yougenerate for youModelsModelsNext Token PredictionText GenerationPatte

2、rn matchingAI that can AI that can chat for youchat for youAI that can AI that can do for youdo for youAssistantsAssistantsPowered by single customized LLMAgentsAgentsPowered by Think-Act-Observe loopFundamental Shift in AI is underwayLLMInput PromptOutputThink&Planhink&PlanObservebserveAct ct Input

3、:Complex TaskOutputFeed forward SystemsFeedback SystemsWith agentic AI,todays front-end systems will become back-end systems.higher quality in work completed with AI tools1 decrease in time to complete tasks when using AI tools140%more tasks completed in total with the help of AI tools112%25%By 2028

4、,1/3 of interactions with generative AI will use autonomous agents for task completion2AI agents will drive the next waveof knowledge worker productivityAn AI agent is an application that can act autonomously to understand,plan,and execute a specific task.AI agents use LLMs to reason and can interfa

5、ce with tools,other models,and other IT systems to fulfill user goals.What is an AI agent?6An AI agent is an autonomous system that can use tools and collaborate with other agents to plan and act on tasks.After it acts,the agent reflects on the results of its actions,learning iteratively and refinin

6、g its approach to better align with its defined objectives.AI AgentsInput(Conversational UI or APP/Business Process)Output of executionThe Orchestrator Agent is a multi-agent,multi-tool supervisor,router,and planner,facilitating complex task executionIBM Data Platform|2025 IBM CorporationAgentic AI