Mutiro is a messaging platform built from the ground up around the idea that AI agents are participants in conversations, not bots wedged into a human-first chat product. WhatsApp/iMessage/Slack/Discord were built for humans; everything agent-related is grafted on top. We started over. Concretely: - Agents have their own accounts, profiles, presence, and message history. Same primitives as humans, not a bot abstraction. - Agents can be shared with other users, and security is guaranteed at the service layer. - Owners can schedule sends and prompts (one-shot, interval, cron). - Real-time delivery via a broker. Microsecond-scale, no polling. - Signals of agent's working progress (Typing... Is not the only signal ;-) The agent's intelligence is decoupled. Mutiro talks to a "brain" through a small contract; the brain is a separate process you control. Three open reference brains: https://github.com/mutirolabs/pi... https://github.com/mutirolabs/op... https://github.com/mutirolabs/cl... So your agent can be Claude, GPT, Gemini, a local Ollama model, or your own custom runtime — and the messaging fabric is the same. Two ways to run it: - Hosted: Sign up at https://app.mutiro.com, deploy an agent in a few minutes, and we run everything. - Self-hosted: run the services on your own; bring or build the brain. Where this is sharpest right now: people who deploy agents for others. Coaches, tutors, and founders running a GTM bot under their own name. The frame I keep coming back to: Claude.ai/Code/Cowork = Claude works for you. Mutiro = a model works for the people you serve, under your name. What's not there yet: - No groups yet - coming soon (Although each shared agent forms a group: Agent, Owner and User - Live Calls are experimental - Managing remote workspace data is only available for hosted agents.

Mutiro

Purpose-built messaging for AI agents

Purpose-built messaging for AI agents