Togo is product analytics with an AI analyst. You send events from your app with a small SDK, and you ask questions about them — in the dashboard, or from your coding agent through Togo’s MCP server.
Choose how to install
- Let your AI agent do it. Paste one line into Claude Code, Cursor, Lovable or any agent that can edit your code: Install Togo with your AI agent. This is the fastest path, and the one we test hardest.
- Install it yourself, by framework: Next.js, React with Vite, plain HTML, or Node.js and serverless functions for events your server sends.
- Already on PostHog, Mixpanel or Amplitude? Keep it, and add Togo beside it: Run Togo alongside another analytics tool.
Whichever you choose, finish with Check that events arrive. It takes seconds with a test key.
Two things to know first
- Your write keys are public. A project has a test key (
togo_pkt_…) for development and previews and a live key (togo_pkl_…) for production. Both go in your frontend bundle and can only send events. Togo never needs a secret key from you. - Live data is a few hours behind. Events sent with the live key reach your reports after the next processing run. Events sent with the test key show up within seconds, but only in the test views. When your data shows up explains why.
The rest of the docs
- How Togo works — projects, event governance, events and their names, people and identity, sessions, and what never goes in an event.
- Connect your AI — the MCP server, for Claude Code and other agents.
- Guides — one task a page: identify people, set profile properties, connect a consent banner, send events from your server.
- Reference — every SDK method and option, every MCP tool, every rejection code.
- Troubleshooting — start here when events do not show up.
Every page here has a Markdown version at its URL plus .md — this one is
/docs.md — and www.togohq.ai/llms.txt maps the whole site for an agent.