AI
Boost.space brings AI directly to your data: the built-in Agent Chat operates your system in plain language, and AI features enrich, transform, and validate the data itself. Connecting an external agent — Claude, ChatGPT, Cursor, or your own — is covered in MCP.
AI always acts within your permissions — it can read, write, and query only the data you can, nothing more.
Start here: the Agent Chat — the agent chat built into the app, zero setup. Bringing your own agent instead, or building for agents? That lives in the MCP tab, starting with Boost.space for AI agents.
AI inside your data
| Feature | What it does |
|---|---|
| Agent Chat | The built-in agent chat — ask, act, and build automations in plain language |
| Vector database | Embed spaces so agents and search match records by meaning |
| AI fields | Fields whose values are generated or enriched from a prompt |
| AI text manipulation | Shorten, extend, translate, and rephrase text inline |
| AI data transformation | Clean and standardize a whole column with AI |
| AI data validation | Check a column against a rule and surface what fails |
AI features run on AI credits.
Bring your own agent
External agents connect over the Remote MCP Server and get the same capabilities as Agent Chat — search and write records, build structure, run AI operations — within the connected user's permissions. Start with Connect via MCP; per-client guides live in MCP clients.
How they fit together
Your data lives in Boost.space. AI features act on it from the inside; MCP lets outside agents read and write it too — both within your permission boundaries. The same operations are available to your own code over the REST API and SDKs.
