Supavec - The open-source alternative to Carbon.ai. Build powerful RAG applications with any data source, at any scale.
⚡ OSS RAG-as-a-Service — spin up vector search + chat API in <5 min
📈 630▲ on Product Hunt · 620⭐ on GitHub
Cloud version
Our simple API endpoints ✨
Related repositories
Built with
API docs
https://github.com/taishikato/supavec/blob/main/packages/api/README.md
Architecture & Design Philosophy
Scalable Multi-Tenant Design
- Row-Level Security (RLS) for team-level data isolation
- Usage‑based billing: Free 100 → Basic 750 → Ent 5000 req/mo
- Batch embeddings: OpenAI cost -65 %
Vector Search Optimisation
- Configurable chunk & overlap (+12 pts recall)
- Hybrid filter (file_id + cosine) P95 210 ms
Performance & Observability
Real‑time Analytics
- PostHog integration product & API events tracked (11 schema)
- Request‑level tracing with unique IDs for rapid debugging
- Async usage logging non‑blocking main flow
API Performance
- Streaming or standard responses selectable per request
- 100‑doc batch embedding reduces latency + cost
- Background processing keeps critical path lean
// Example: choose streaming or standard chat responses
const response = await fetch('/chat', {
method: 'POST',
body: JSON.stringify({
query: 'What is this document about?',
file_ids: ['uuid-here'],
stream: true // set false for JSON
})
});
Developer Experience
- Visual debugging: live embedding preview in chat UI
- Progressive disclosure: guided onboarding & contextual states
- API-first design: REST endpoints + comprehensive errors
- Redis-backed rate limiting: sliding-window strategy
Getting Started
Install dependencies
bun i
Run the development server
bun dev
Run the web server
cd apps/web && bun run dev
Run the API server
cd packages/api && bun run dev