JP Morgan Deploys 'Ask David' Multi-Agent System for Investment Research
JP Morgan's investment research team has deployed "Ask David," a multi-agent system built on the pattern now described as "showing up everywhere": a supervisor agent orchestrates specialized subagents for retrieval, structured data, and analytics; an LLM-as-judge reflection node validates output before delivery; and human-in-the-loop closes the final accuracy gap.
Why It Matters
A Tier-1 investment bank running multi-agent research AI in production — with HITL as a design choice, not a concession — validates the architecture as enterprise-ready. The supervisor+specialists+judge+HITL pattern is becoming the de facto enterprise agent template.