Bricktelligence

The problem

Turning a flood of community content into a curated, personality-driven feed required memory, judgement, and a publishing pipeline — autonomously, every day.

Architecture

140K lines of TypeScript across Next.js + React Native on Supabase. A memory-driven RAG persona, an autonomous scoring/curation agent on a daily cron via the Anthropic API, a 13-source ingestion pipeline, and Meta Graph API publishing.

What I shipped

• ▹Memory-driven RAG persona
• ▹Autonomous daily content-scoring & curation agent (Anthropic API)
• ▹13-source ingestion pipeline
• ▹Meta Graph API publishing pipeline
• ▹Shipped to iOS