Meet Anton
Anton is an open-source AI coworker that can execute tasks, connect to tools and data, remember lessons, and improve its workflows over time.
What Anton does
You describe the outcome; Anton figures out the steps. It writes and runs code in an isolated workspace, connects to your databases and apps, searches the web, and delivers the result — a report, a cleaned inbox, a live dashboard, a working integration.
- Hand off real tasks — automated reporting, email cleanup, calendar management, research and synthesis, workflow automation.
- Connect your data and apps — Postgres, BigQuery, Snowflake, Gmail, Salesforce, HubSpot, and more, with credentials kept in a local vault that is never exposed to the model. See Connect things.
- Teach it your way of working — Anton saves lessons, rules, and reusable skills as it works, and gets better at your workflows over time. See Teach Anton.
- Use any model — Minds, Anthropic, OpenAI, Gemini, or any OpenAI-compatible endpoint, including local ones. See Pick a provider.
Anton is not a coding copilot. It writes code as a means to deliver an outcome, not as the product. And it is not a chatbot — it takes action, remembers what it learned, and improves.
Where to go next
- Quickstart — install, run your first task, and watch Anton learn from it, in about five minutes.
- Use Anton — the terminal interface, sessions, and workspaces.
- Reference — every CLI and slash command.
Curious how Anton thinks? Read the architecture overview — the contributor guide covers the brain-inspired memory systems, the scratchpad runtime, and how to add tools and data sources.