构建知识代理以自动化文档工作流程.pdf

1、Building Knowledge Agents to Automate Document WorkflowsJerry LiuCEO,LlamaIndexJune 10-11,2025AI Agents can“Automate Knowledge Work”2A big promise of AI agents is making knowledge workers more efficient:Lower cost,save time More data,better decision-makingEvery b2b SaaS is talking about it.But does“

2、knowledge work automation”just mean RAG chatbots?90%of Enterprise Data Lives in Documents*3Humans historically needed to read/write these documents.AI agents can automate this end-to-end workflow but they need access to the right context and architecture to be useful!Agent*IDC.(2023).Untapped Value:

3、What Every Executive Needs to Know About Unstructured Data White paper.IDC https:/ in the Tools and ArchitectureToolsSurface relevant context.Let agent take external actions.Connect through MCP/A2AArchitectureEncode business logic through agent reasoning.+Necessary ingredients for automationHelping

4、Humans with Document WorkAssistive AgentHelp humans get more information,faster.Automation AgentAutomate routine tasks in the background,take actions on behalf of the user.A Lot of Knowledge Work is Rooted in Documents5Knowledge WorkerDocument Research AgentsAutomation AgentsFinancial AnalystsPerfor

5、m due diligence on public/private reports,generate memosExtract and normalize financial metrics for equity research.Engineering+R&DLook up engineering docs,designs,and specifications,generate PRDsExtract and validate technical specifications from complex datasheets.CliniciansSearch clinical guidelin

6、es and medical research papers,generate summariesProcess patient records(EHRs)to ensure data consistency.Customer SupportRetrieve product documentation,troubleshooting FAQ.Extract and classify returns and support tickets.Administrative OperationsAssistants over onboarding materials,review legal cont