Sugabots is a self-hosted workspace where people and AI agents work together. Create agents, organise them into shared pods, and run conversations that can use built-in tools and MCP connections.
Everything stays in your own database. Bots use the model providers you choose, whether that's Anthropic, OpenAI or a model server on your own network.
With Docker and Docker Compose installed, run this on macOS or Linux:
curl -fsSL https://sugabots.ai/install.sh | shOr on Windows, in PowerShell:
irm https://sugabots.ai/install.ps1 | iexIt sets Sugabots up in a new sugabots folder and starts it. Then open http://localhost:3000 and create your account. You can read more details in the
Quickstart guide.
To use Sugabots with other people, run it somewhere they can reach. Deploy with Docker walks through a server of your own or a hosting platform.
A pod is a shared space for a group of people and their bots. Everyone in it sees the same chats, results and approvals, so nobody has to forward an answer or explain what a bot already did.
When a bot needs help, it asks another bot in its pod. The exchange shows up as a thread under its message, so you can see exactly who said what.
Connect Anthropic, OpenAI, any OpenAI- or Anthropic-compatible service, or a model running on your own machine. Give each bot the model that suits its job: a clever one for research, a cheap one for sorting the inbox, a local one for private notes.
Connect Linear, Notion, Stripe, Sentry, Jira, any MCP server or a plain HTTP API once, and let your bots use it to do work. Choose which tools run straight away and which wait for someone to say yes.
Give a bot a job to do on a schedule, or whenever a webhook or a GitHub event calls it. It posts what it did in its chat, where the team can pick it up. Try telling a bot "check in with me every morning at 9".
Bots can runs code, use software or browse the web in their own isolated environment.
- GitHub: bots read a pod's repositories, push branches, and open pull requests once someone allows it.
- Artifacts: files and outputs bots make are saved for the team to open and review.
- Usage and costs: see what every bot, model and team costs, day by day.
- Web search: bots search the web and read pages, through Exa, Brave Search, Tavily or your own SearXNG server.
- Roles: owners, admins, members and viewers, with pods deciding who sees what.
- Personal pod: everyone gets a private pod and assistant that nobody else can see into, admins included.
- Bots set their own schedules: ask a bot to "check in with me every morning at 9" and it sets up the routine.
- Signed webhooks: routines check calls from GitHub, Stripe, Slack and Linear, or any sender using HMAC, bearer, API key or basic auth.
- Ask questions and get help on the Sugabots Discord.
- Read the docs at sugabots.ai/docs.
- Report bugs and request features in GitHub issues.
To work on Sugabots, see CONTRIBUTING.md.
The software source code in this repository is licensed under the MIT License unless otherwise stated. See the asset exceptions below and the package-level licensing notices.
Sugabots is a product of Nitric Group Inc.
The Sugabots name, logo, brand identity, and supplied brand assets (including illustrations and avatars) are not licensed under the MIT License. See the notices for logos and avatars and website brand and marketing assets.
The avatar geometry, colour definitions, and rendering code remain MIT licensed.
Forks and derivative products may use the MIT-licensed software, but should use their own name and branding. Unofficial forks and derivative products must not imply affiliation with, endorsement by, or maintenance by Nitric Group Inc. or Sugabots.
Third-party materials retain their respective licenses and notices, including the provider logos.