FollowAI

Follow any artist, team, league or event. FollowAI scans the web with AI and sends only what's new. One calm timeline instead of ten apps and endless scrolling.

Comment

Inspiration

I follow a handful of things closely: a couple of bands, a hockey league, an artist or two. Keeping up meant checking five apps, three websites and a feed built to keep me scrolling. News I actually cared about got lost between things I didn't. I wanted the opposite: tell an app what I care about once, and have it tell me when something happens, and nothing else.

What it does

  • Follow anything. Search for a person, band, team, league or tournament, or pick from a shared pool of subjects other people already follow.
  • AI finds the news. FollowAI checks every followed subject on the live web and writes a short, sourced summary of what's new. Nothing is made up from memory: every update links back to a page that was actually read.
  • One calm timeline. Each follow gets its own timeline, and all news lands in one Updates tab.
  • Stars pay for the checks. Every new user gets free Stars. Following a new subject costs 10★ and fetches its news straight away. Joining a subject someone already follows costs 1★, because its timeline is already there. Stars are available as one-off packs or as a subscription, both handled through RevenueCat.

How we built it

  • App: Flutter for iOS and Android phones, with Riverpod for state and go_router for navigation.
  • Backend: Supabase: Postgres, Edge Functions.
  • AI: LLM model with web search tool.
  • Push: Firebase Cloud Messaging.
  • Payments: RevenueCat for Star packs and stars subscriptions.
  • Sign-in: Apple and Google sign-in (with a nonce), plus email.
  • Process: SDD (OpenSpec) + Claude Code as a pair programmer.

Challenges we ran into

  • The business model. AI tokens aren’t cheap. It was hard to find a good monetization strategy, and I’m still not sure I picked the right one.
  • Empty on day one. .

Accomplishments that we're proud of

  • Shipped and approved on the App Store.
  • Payment model
  • Every update in the app links to its source. No hallucinated news.
  • A shared subject pool, so popular follows get cheaper to run as more people join.
  • The whole backend runs on Supabase with no separate servers to maintain.

What we learned

  • Price on the same thing your costs are measured on.
  • RevenueCat made consumables plus a subscription surprisingly painless on both stores, and its webhooks made a server-side balance easy to trust.
  • Writing the spec first made changes that looked scary, like replacing the whole billing model, into a checklist.
  • RevenueCat MCP is a great tool. Thanks.

What's next for FollowAI

  • More kinds of things to follow, like podcasts, companies, games and TV shows.
  • Smarter checks: check busy subjects more often and quiet ones less, so Stars go further.
  • Richer updates: dates you can add to your calendar (match days, releases, tour dates).
  • Share a follow with a friend, and gift Stars.
  • More languages, starting with news written in the subject's own language.

Built With

Share this project:

Updates

Submission history