AIDevToolsAPICollaboration

Multi-Agent Conversation Bridge

Let different people's AI agents negotiate tasks in controlled rooms

Problem
Experiments already let agents chat with other people's agents, showing multi-agent interop is moving from toy to real need. But there is no self-serve, indie-friendly "agent lobby" that isolates each owner's context while recording the negotiation.
Solution
A lightweight agent gateway: dedicated rooms per user/team, with agents connecting via simple HTTP/WebSocket, supporting capability declarations, task negotiation, and result callbacks; every conversation is auditable and replayable, with configurable redaction before outbound messages.
Users
Indie developers running multiple AI agents, automation enthusiasts, and small AI product teams
MVP
Agent registration and capability declaration; rooms and invite codes; WebSocket-based message relay; conversation logs and replay; basic redaction rules
Revenue
Free tier plus $9–$29/month hosted rooms subscription; usage-based API billing
Why now
The number of agents per person and small team is growing fast, and cross-agent collaboration still lacks a lightweight standard; an indie developer can claim the developer experience with a minimal protocol

Source discussions · 2

  1. Hacker NewsShow HN: OnPanda – Steer LLMs and agents at the token level2↑ 0
  2. Hacker NewsShow HN: Yello. Let your agents chat with other people's agents2↑ 2
← Back to list