Kronvex started as a personal frustration — AI agents that forgot everything between sessions.
My name is Baptiste. I'm a solo developer based in France, and I built Kronvex because I kept running into the same wall: every AI agent I shipped was stateless. The moment a session ended, everything the agent learned — user preferences, past decisions, context — disappeared. Gone.
I started exploring solutions. Most required spinning up a dedicated vector database, writing custom embedding pipelines, and figuring out similarity search from scratch. It was a two-week detour just to get a prototype working. And then I had to maintain it.
So I built the layer I wished existed. FastAPI backend, PostgreSQL with pgvector for cosine similarity search, OpenAI embeddings, hosted entirely in the EU on Supabase Frankfurt. Three endpoints: remember, recall, inject-context. That's it.
I'm building Kronvex in public. No VC funding, no enterprise sales team. Just a developer who got tired of re-solving the same problem and turned the solution into a product. EU-hosted by principle, not just compliance.
"Build the memory layer that makes AI agents actually useful — persistent, semantic, GDPR-native, production-ready."
REST API with OpenAPI docs. Python and Node SDKs. No vendor lock-in — your memories are yours, exportable anytime.
All data lives in Frankfurt (Supabase EU cluster). GDPR isn't an afterthought — it's baked into the architecture from day one.
Fixed monthly plans. No per-call billing surprises. The price you see is the price you pay — clear limits, no hidden multipliers.
The stack is intentionally boring. Proven tools, EU infrastructure, no exotic dependencies.
No credit card. Your first memory stored in under 5 minutes.
Try Kronvex free →100 free memories. No credit card required.
Already have an account? Sign in →