Your dev team
never sleeps.
Talk to AI coding agents through Discord voice calls or Slack channels. Give them work and disconnect. They keep shipping.
Momentum is
everything.
I get most of my ideas while driving. The problem is I always have to wait until I get home to act on them, or write them down and hope I remember the context later.
Every hour between having an idea and starting to build it is wasted time. That's the time value of momentum. Shipmates lets you call your AI dev team on Discord or Slack the moment inspiration hits. Describe what you want built, hang up, and it's already in progress before you pull into the driveway.
Four moving pieces.
Voice-first development
Join a Discord voice channel, tell your agents what to build, and disconnect. They keep working independently, in isolated containers, against your real repos.
Real code, not demos
Agents run Claude Code, Codex, or Gemini CLI inside Docker containers. They write code, run tests, push branches, and open PRs on your actual codebase.
Async clarifications
When agents get stuck, they message you on Discord or Slack. Answer when you can. They pick back up automatically. No babysitting required.
Your keys, your models
Bring your own API key for Anthropic, OpenAI, Google, or 5 other providers. Assign different models to different agents. Mix and match however you want.
Name them. Role them. Ship.
Each agent gets a name, personality, voice, and scoped access. They stay in their lane so they don't step on each other's changes.
Alex
Captain
Sam
Engineer
Jordan
Lookout
Casey
Navigator
Morgan
First Mate
Scoped access
Assign agents to specific repos, directories, or languages. Your frontend agent only touches the UI. Your backend agent stays in the API layer.
Distinct voices
Pick from realistic voices powered by ElevenLabs or use free local TTS via Piper. You can tell who's talking without looking at a screen.
Ready to set sail?
Clone the repo, run the setup script, and start talking to your agents.