FounderOS is a decision memory system for startups powered by Qwen Cloud. It solves a critical problem: founders repeatedly lose decisions, lessons, and context and end up repeating costly mistakes.
How it works: Every message a founder sends is automatically classified as a decision, question, or thought using qwen-plus. Decisions are embedded using Qwen text-embedding-v3 and stored in a pgvector database with an importance score. When a founder asks a question, the memory engine retrieves the top most relevant memories using vector similarity search combined with importance scoring.
Memory Intelligence:
- Importance scoring: memories gain score on every recall, decay after 30 days of no access
- Conflict detection: before storing, Qwen checks if the new memory contradicts existing ones
- Proactive contradiction warnings: before answering any message, the agent checks for contradictions and warns the founder proactively
- Duplicate detection: exact duplicates are never stored twice
Built with Qwen Cloud:
- qwen-plus: intent classification, conflict detection, contradiction checking, response generation
- text-embedding-v3: semantic vector embeddings for all memory storage and retrieval
Deployment:
- Frontend + Backend: Alibaba Cloud ECS (Singapore, ap-southeast-1)
- Database: Supabase PostgreSQL with pgvector
- Auth: Supabase Auth
Built With
- alibaba-cloud-ecs
- next.js
- pgvector
- qwen-cloud
- supabase
Log in or sign up for Devpost to join the conversation.