I built a multi-agent assistant that sends each request to the right specialist. You type a request. A router…
I built a multi-agent assistant that sends each request to the right specialist. You type a request. A router agent reads it and hands it off to one of three agents: - Writer: drafts emails, posts, and articles - Reviewer: checks text you've already written and gives feedback - Researcher: explains a topic with background and key points Every reply shows which agent answered, and the answer streams in as it is written. How it's built: - OpenAI Agents SDK (the framework) for agents and handoffs - Gemini as the model, used through its OpenAI-compatible API - Chainlit for the chat interface - Deployed on Render Limitations: the Researcher has no web search, so it answers from the model's own knowledge. It runs on a free instance, so the first request after a period of inactivity can take up to a minute to wake up. The demo also uses a free API quota, so it may stop responding if the limit is reached. The recording above shows it handling one request for each agent. Live demo: https://multiagent-chainlit.onrender.com/ Code: https://lnkd.in/dsQj9nQk #AI #AIAgents #Python #Chainlit #Gemini