Bee by Heossi
The Progressive Quantum-Native Intelligence Engine
Start new thread

Bee for Developers - The quantum-native LLM API — OpenAI-compatible

by•
Bee's developer surface is live: a drop-in OpenAI-compatible /chat/completions endpoint, official Python (bee-sdk) and TypeScript (/bee) SDKs, a CLI for scripting, and a hosted MCP server for Claude, Cursor, Continue, and Zed. Point your existing OpenAI code at Bee and it works — no rewrites. Tiered models, tenant-isolated retrieval, post-quantum security paths, and base models never trained on your data. Built by HEOSSI in Singapore.

Add a comment

Replies

Best
Maker
📌
Hi Product Hunt — Christopher here, founder of HEOSSI. When we built Bee's API, the design goal was simple: if you've written code against OpenAI, you should be able to point it at Bee and have it just work. Same /chat/completions shape, official Python and TypeScript SDKs, a CLI for scripting, and an MCP server so Bee drops straight into Claude, Cursor, Continue, and Zed. What's underneath is the part we care about most. Tenant-isolated retrieval. Base models never trained on your data. Post-quantum security paths for teams that need them. And a tiered model ladder so you can start on the free tier and scale to governed or sovereign deployment without changing your integration. $10 in API credits for the PH community to try it. Docs are at bee.heossi.com/docs — I'll be here all day answering anything technical. What's the one thing that would make you switch LLM providers?

A native SDK for at least one mobile framework would make this way more approachable. Most teams experimenting with post-quantum secure LLM calls will want to prototype on-device before wiring up sovereign backends, and right now it sounds like everything routes through the API.

Maker

 That's a very good suggestion. We already build Bee's own iOS and Android app with React Native, and our TypeScript client is fetch-based — but we haven't yet packaged and documented a proper mobile-framework integration for external developers. A tested React Native starter, including the right server-side API-key pattern, would be the natural first step. Thanks for calling this out — noting it as a real gap.