Multi-Agent Workflow

Watch 3 specialised agents collaborate in real-time: Researcher → Critic → Writer.

🔍
Researcher Agent
Retrieves facts via BM25 RAG · outputs bullet-point evidence
Idle
Waiting…
🧐
Critic Agent
Evaluates findings · flags gaps · recommends emphasis
Idle
Waiting…
✍️
Writer Agent
Synthesises research + critique into final polished answer
Idle
Waiting…

✅ Final Answer (Writer Agent)

Query

How it works

1. Researcher — receives your query and retrieves the most relevant knowledge using BM25 keyword scoring over the portfolio database. Outputs structured evidence.


2. Critic — reads the Researcher's findings and identifies gaps, incorrect emphasis, or missing context. Provides feedback to improve quality.


3. Writer — synthesises both outputs into a polished, professional final answer — taking the best of research + critique.


Pattern: Researcher → Critic → Writer (sequential, no library dependencies)