Airtable is a database.
We're a CRM with native MCP.
Airtable is a great database. We use it ourselves for inventory, content calendars, ops trackers. But running a CRM out of it means designing your own schema, wiring your own formulas, and — if you want AI to drive any of it — writing your own MCP server on top of the REST API. OneHub360 ships the CRM and the MCP server. Bring Claude or ChatGPT to the data instead of writing glue code to it.
Side by side
| Airtable as CRM | OneHub360 MCP | |
|---|---|---|
| Data model | Generic relational base — you design the schema | Opinionated CRM schema: contacts, deals, tasks, proposals, invoices, calls |
| Deal pipeline | A view on a table you set up yourself | Built-in pipeline with stages, weighted forecast, age-in-stage, drag-to-move |
| Contact records | Linked records to a Contacts table you build | First-class Contact entity with email/phone validation, dedupe, owner, tags, activity timeline |
| Proposal generation | Build a doc-merge automation yourself | Native proposal builder with line items, signature workflow, client portal |
| Invoice generation | Build it with extensions or third-party Apps | Native invoices with Stripe payment links, status tracking, dunning |
| Email integration | Sync via Zapier or scripting block | Two-way Gmail/IMAP sync, threaded inbox tied to contacts and deals |
| MCP server | None. REST API + Scripting Block — bring your own MCP wrapper. | Hosted at /api/mcp — 28 tools (19 read + 9 write incl. destructive) |
| Formulas vs CRM logic | Formula fields per column — you wire deal value × probability yourself | CRM logic baked in: weighted pipeline, win rate, conversion, ROI |
| API-only vs MCP-native | REST API. Agent has to know your schema, your bases, your view IDs. | Tools described to the agent in MCP — agent introspects and acts |
| Build-your-own integration | Required for any AI agent workflow | Out of the box. Connect Claude/ChatGPT/Cursor with one config snippet |
| AI agent support | Airtable AI summarizes records. No agentic write path. | Claude/ChatGPT/Cursor read pipeline, create deals, send proposals, log calls |
| Plan-tier gating | Per-base permissions, manual API key scoping | Built-in: Starter ($97) read, Growth ($197) read+write, Pro ($397) full |
| Audit log | Revision history per record | ApiAuditEvent on every MCP call — actor, tool, args, result, timestamp |
| Mobile app | Yes — for editing records | Yes — for running pipeline, replying to inbox, taking calls |
| White-label | Interface designer with custom branding | Pro tier — your domain, your logo, full client portal |
| Pricing | $20/user/month (Team plan), then per-seat | $97/month flat, unlimited team |
Why this matters
Airtable is the best general-purpose database
Truly. For anything that's not a CRM, it's usually the right call: project trackers, inventory, editorial calendars, applicant tracking. The flexibility is the feature. But that flexibility means there's no shared shape that an AI agent can know in advance — every Airtable base is a snowflake.
REST API isn't MCP
Airtable has a great REST API. That's not the same as a hosted MCP server. To get Claude or ChatGPT to drive Airtable, you're writing — and hosting, and securing, and rate-limiting — your own MCP wrapper around their API. We've already done that work for the CRM use case. 28 tools, plan-tier filtered, audit-logged, tenant-scoped per API key. Six lines of config and you're done.