Your semantic layer, in any AI tool you use.
Truefold exposes your reconciled metrics through an MCP server. Point Claude, ChatGPT, Cursor, or your own agents at it: they query your layer with full source citations. You are not locked into our chat UI.
Your semantic layer, in any AI tool you use.
Truefold exposes your reconciled metrics through an MCP server (Model Context Protocol). Point Claude, ChatGPT, Cursor, or your own agents at it: they query your layer with full source citations. You're not locked into our chat UI.
- Compatible with Claude Desktop, ChatGPT, Cursor, and any MCP-aware client
- Public hosted endpoint or self-hosted (Docker, ~1GB RAM)
- API-key auth, scoped per-team
- Queries from your AI tool stay between you and the tool; Truefold never sees them
- Export the full semantic layer as JSON or dbt-compatible YAML for your own stack
Your data team writes a Cursor agent that asks Truefold "any campaign with ROAS below 1.5× for two consecutive weeks?" and posts to Slack every Monday. Same semantic layer, your team's tools, your team's workflow.
Read-only on every source.
Truefold pulls; it never writes. No platform changes, no risk of clobbering your data.
You own everything.
Metric definitions, customer segments, queries: all exportable. Leave any time with what you built.
We don't train on your data.
Your business data never enters a model training set. Single-tenant by default.
Ready to see your real numbers?
Beta access is rolling out to ecommerce brands a few at a time. Tell us about your stack.