Product
Ground is the context layer for teams shipping AI features: index once, query everywhere (MCP, API, dashboard, public agents). Less prompt engineering, more shipping.
Search that keeps up with your stack
Hybrid keyword + vector retrieval with optional multi-hop expansion. One API surface for repos, docs, and packages—built for agents that need the right file, not a wall of links.
Answers tied to real code
Every result includes path, score, and source so you jump straight into the editor. No “trust me” answers—your team verifies in one click.
Your sources, synced once
Index Git repos, docs sites, PDFs, and npm/PyPI packages. Add a source, run sync, and every agent gets the same up-to-date context.
On-demand package search
Probe any public package before you add it to your stack. Fetch, chunk, and rank in one request—ideal when you are moving fast on a new dependency.
MCP for IDEs and agents
One npx command wires Ground into Cursor, Claude Desktop, or any MCP host. Same tools your dashboard uses, in the editor where you ship.
Full MCP guide: docs.trygroundai.com/guides/mcp
HTTP API and SDKs
REST endpoints for search, chat, sources, and jobs. TypeScript and Python clients for when you productize context into your own agent.
Public docs agents
Launch a branded search page at /agent/your-slug scoped to your docs or repo. Share with users, prospects, or your YC demo—live in minutes.
Rover (deep GitHub)
Autonomous GitHub exploration when a question needs listing, reading, and globbing across repos—separate mode from everyday search.
Try it now
Same retrieval stack as production—pick a prompt and see how fast context comes back.