AI assistant & RAG demo suite
A suite of production-style AI demos: a streaming chat assistant (Next.js + Vercel AI SDK), two RAG pipelines (FastAPI + Pinecone, and Supabase / pgvector), and a tool-using research agent — each showing grounded, streaming AI, not slideware.
See it in action
What we solved
Prospects want to see AI that actually works — grounded, streaming, and tool-using — before they trust it in production.
How it was built
Streaming chat assistant with the Vercel AI SDK on Next.js
RAG over documents with FastAPI + Pinecone, and a Supabase / pgvector variant
A research agent with tool use and streamed reasoning steps
The result
A one-click way for clients to try real, grounded AI
Streaming, RAG, and agents demonstrated in one place
Reusable patterns that carry straight into client work
- Streaming LLM chat (Vercel AI SDK)
- RAG: Pinecone + Supabase / pgvector
- Tool-using research agent
Have a similar project?
Tell me about it and I'll respond within a few hours.