Inspiration
Creators with millions of followers have zero intelligence about who their actual superfans are. No tool tells you who comments on every post, shares your content across platforms, or mentions you on LinkedIn and X. We wanted to fix that.
What it does
FanIQ is an autonomous fan intelligence platform. A creator enters their handle, and FanIQ's Tinyfish web agent crawls X and LinkedIn in real-time — no API keys needed — extracting every person who has engaged with them. Those people become enriched fan profiles scored and ranked live in Redis. Then the creator can open a voice call and ask "who are my top fans?" or pick a fan from the leaderboard and have an actual voice conversation with an AI persona built from that fan's real comments. Every fan profile is published to cited.md via Senso so other agents can discover and pay for the intelligence autonomously.
How we built it
- Tinyfish web agent crawls X and LinkedIn using natural language goals
- Redis sorted sets power the real-time fan leaderboard with sub-millisecond scoring updates
- Vapi custom LLM endpoint serves two voice modes: intelligence (ask about your fans) and persona (talk as a fan built from their real engagement data)
- *Ghost * stores persistent fan profiles as structured member records
- FastAPI backend with async SSE streaming so the UI updates live as fans are discovered
Accomplishments that we're proud of
The Vapi persona mode actually works; a creator can select a fan from the leaderboard, start a voice call, and have a real conversation with an AI that speaks using that fan's actual words and opinions. The Tinyfish integration crawls live social media with zero platform API keys.
What we learned
Vapi's custom LLM pattern is extremely powerful but requires the backend to be fast, any latency over ~2s breaks the voice illusion. Redis sorted sets are the perfect data structure for real-time leaderboardst. Tinyfish's goal-based web agent removes an entire class of infrastructure problems (proxies, selectors, anti-bot) that would have taken days to solve otherwise.
What's next for FanIQ
Expand to Instagram, YouTube, and TikTok. Add a continuous monitoring mode that alerts creators when a new superfan emerges. Build the full x402 payment rail so fan intelligence reports earn micro-payments autonomously. Let creators send personalized messages to their top fans directly from the dashboard, powered by the same persona data that drives the voice calls.
Built With
- redis
- tinyfish
- vapi
Log in or sign up for Devpost to join the conversation.