Engineering Intelligence through Agent Interaction
summary
Transforms RPA into agentic process automation by combining GPT‑4 Vision and vector/semantic indexing into chained, policy‑based agents that keep source references to avoid hallucinations. A working file‑processing system for a pump manufacturer handles hundreds of gigabytes of heterogeneous files, uses compartmentalized memory and limited‑action agents, and preserves audit trails plus transient handling of private data. Solution scales in practice but depends on third‑party model versions and needs ownership mapping and guardrails to prevent emergent or manipulative behavior. policy‑based agent
AI-written summary · about
chapters (10)
- 0:00Engineering intelligence intro
- 2:12Intelligence via agent interaction
- 3:59Bottom-up AI automation
- 8:09Avoiding hallucinations in reports
- 10:57Auditing client processes
- 13:26Compartmentalizing agent communication
- 17:59Hierarchical associative memory
- 20:28Distributed intelligence systems
- 23:21Superintelligence challenges
- 24:05Q&A
related talks
26mBuilding Agentic Workflows: Patterns for Orchestrating Intelligent Systems.
22mAgentic AI for Engineering Leaders | Strategy & Execution Guide
28mAgentic AI
43mAI Procurement Agents | The Future of Intelligent Procurement
46mIntelligence as Infrastructure: What Developers Must Learn to Build with AI
29mAPI Integration augmented by GenAI
36mBuilding Agentic AI Assistants: Lessons from productionising agents in Preclinical R&D
20mReimagining InnerSource for the Agentic AI Era