Back to home
AI · SocialLive

Bloom

An AI-native content platform — feed, music, short films.

Role
End-to-end development — architecture, backend, front-end, DevOps.
Timeline
6 months from blank repo to closed beta.
Stack
AISocialNext.js
Context

Bloom is a social platform built around AI-generated content. The feed mixes short films, music tracks and visual posts, all produced or augmented with generative tools. Users discover, react, remix.

Bloom started as a thesis on what social feels like when the cost of creation drops to near-zero. The team wanted a product that could host AI films, AI music and traditional posts side-by-side, with a feed that felt curated rather than chaotic.

The problem

Generative content is abundant. A feed full of AI output without ranking is noise. The challenge was building a discovery layer that surfaces taste — and a creation flow that's fast enough for daily posting.

Approach

Unified post model

Films, music and posts share one schema, with media-specific renderers. Adding new content types is a one-day job.

Recommendation feed

Lightweight scoring (recency, engagement, taste) tuned to surface variety, not just popularity. Cold-start handled by curated starter packs.

AI creation flow

Direct integrations with video and music generation APIs, queued through a job system. Drafts saved automatically.

Mobile-first UX

Swipe navigation, polished media player, optimistic interactions. Built to feel native even on the web.

Outcome
  • Production-ready platform shipped to beta users
  • Average session time above 8 minutes in early testing
  • Content pipeline supporting 3 generative providers behind one API
Next.jsTypeScriptPostgreSQLRedisS3Stripe

Something similar in mind?

Tell me where you stand — I get back within 24h.