Featured
Research Crew
A multi-agent system that conducts in-depth research on any topic using specialized agents (researcher, critic, summarizer).
- Multi-Agent
- AI Agents
- LangChain
- CrewAI
- Updated Sep 12, 2026
- Experiment
- Type
- Experiment
- License
- MIT
- Language
- Python
- Updated
- Sep 12, 2026
Overview
Research Crew splits a research question across specialised agents: a researcher gathers sources, a critic challenges weak claims, and a summarizer writes the final brief with citations.
The experiment compares a single-agent baseline with the crew on accuracy, depth and cost.
Key Features
- Role-based agents with separate prompts and tools
- Critic loop that sends weak sections back for revision
- Citations for every claim in the final brief
- Cost and token tracking per agent
Tech Stack
LangChain
CrewAI
OpenAI
Python
Project Links
What I Learned
- A dedicated critic catches more errors than asking one agent to self-review.
- Clear hand-off formats matter more than the number of agents.
- Parallel research cuts latency but needs deduplication.
Next Steps
- Add a fact-checking agent with web search
- Benchmark against human-written briefs