Questions,
answered plainly.

What is Unfurl?

A local MCP server that syncs HubSpot and Stripe into a resolved revenue graph, deduped, cross-linked, and annotated with data-health caveats, and serves it to AI agents through six read-only tools.

Which agents does it work with?

Anything that speaks MCP: Claude Code and Claude Desktop, Cursor, and other MCP clients. It runs over stdio.

Does it write to my CRM?

No. Every tool is read-only and the credentials it uses are read-only. If write actions are ever added, they will propose changes and wait for a person to approve them.

Where does my data go?

Into a SQLite file on your machine. Unfurl has no server of its own to send it to.

Which data sources are supported?

HubSpot (companies, contacts, deals, owners) and Stripe (customers, subscriptions). More sources will follow early-access requests.

Do I need a RevOps person or a sales call to set it up?

No. Try it on the demo dataset in two commands, then add a HubSpot token when you are ready.

Does it forecast?

It does not predict. It makes sure the inputs your agent forecasts from are clean: merged duplicates, counted missing amounts, and flagged stale close dates.

Can someone run it for us?

Yes. The RevOps module from Humonix runs on Unfurl: it sets up your stages, scoring, and reports, and keeps them current every week, with every CRM change approved by a person. Details at humonix-web.fly.dev/services/revops.

How much does it cost?

Unfurl is in early access. Team pricing will be published on the Pricing page before general availability.

Ask us directly.