在 Amazon ECS 上构建安全且可扩展的代理 AI 工作流.pdf

1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.C N S 3 4 7Build secure and scalable agentic AI workflows on Amazon ECSDave HorneSenior Partner Solutions ArchitectWorld Wide Public SectorSanjeev GanjihalSenior Spe

2、cialist Solutions ArchitectWorld Wide Specialist Organization 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.The GenesisUserQueryResponseAI AssistantLLMPromptResponseTraining data as of date 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.A simple RAG workflowE

3、nterprise knowledge basePopulateUserAI AssistantQueryGet extra contextContextLLMPrompt+contextResponseResponse(Vector embeddings)2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Autonomous Agent OverviewUserAgentQueryLLMPrompt&tools descriptionUse tools to get more contextTool resu

4、ltFinal responseResponseRepeat as many times as requiredToolCallResultData source 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Characteristics of an Agent 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Multi-agent customer service exampleHow do changes in sh

5、ort-term interest rates influence growth and inflation?What are the subsequent effects on equity and bond market returns?Customer Service AgentSupervisorService Status AgentKnowledge Base AgentRefund AgentSpecialized agentsUser 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Why c

6、ustomers choose ECS for AI workloadsConfigurabilityFull control of your data and infrastructure configurationRun anywhereConsistent deployments across clouds,on premise,and at the edgeSimplicityNo control plane to manage;No workers to manage with FargateCost optimizationControl of environment to dep