MarketSense AI

A team of three AI agents that detects competitor price drops, analyzes the right response, and queues a human-approved pricing decision — collaborating live on Band. Tags: multi-agent band-of-agents ai-ml-api langgraph competitive-intelligence pricing human-in-the-loop enterprise-workflow fastapi e-commerce Long description: The problem. Retail pricing teams can't watch every competitor on every marketplace 24/7. By the time a competitor's price drop is noticed, sales have already leaked. But fully automating price changes is dangerous — it risks margin collapse and brand damage. MarketSense AI solves this with a team of three specialized agents that collaborate on Band, keeping a human in control of the final decision: Scout continuously scans competitor prices and social sentiment, flagging significant drops. Analyst is recruited into the conversation, requests sentiment from Scout (a genuine bidirectional agent exchange), weighs match / undercut / hold strategies against a margin floor, and writes a strategic recommendation. Executive drafts the proposed action and queues it for human approval — nothing executes until a person clicks Approve on the dashboard. Why it's reliable. Agents pass lightweight references over Band while all structured data lives in Postgres, so decisions are auditable and never depend on parsing chat. The whole pipeline — detection → analysis → governed action → Slack alert — runs autonomously in the cloud, ending at a human-in-the-loop gate.