Skip to Main Content

Traction Radar

Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

Demo video · wildanlsfs.github.io/…

Traction Radar is an AI agent that finds where an app's first users will come from. Give it a product URL and it reads the landing page, searches public conversations for people already describing the problem, scores every post for buying intent, and clusters them into pain points in the users' own words. It then plans three 7-day experiments on different channels (Threads posts, helpful replies, intermediary outreach or a small ad test), each with ready drafts and its own tracking link. Nothing is posted without the founder's approval. Clicks, signups and payments are measured with a one-line snippet, and every day the agent decides: keep, stop, needs more data, or ready for ads. Every outcome is stored anonymously in a "traction bank" that the planner reads first, so each experiment makes the next recommendation better. I built it because I'm an indie developer: building my apps took days, but finding the first users took months. Real result: I ran it on my own product, Taarufin. Three Threads posts from its account brought 15 real clicks in under three days (bots filtered); the agent's verdict was "needs more data", and that outcome is the first entry in the traction bank. It has also scanned three more of my products. Live read-only demo: https://d31lz0ilgvjrj6.cloudfront.net/#/demo Try it on your own product (free, no account): https://d31lz0ilgvjrj6.cloudfront.net/#/try

Team