Attio vs Close
Attio and Close are both modern CRMs whose official hosted MCP servers let an agent do real customer-relationship work over OAuth, which makes them genuine substitutes — but they emphasize different strengths. Attio's server connects an agent to a flexible, data-model-first CRM: search and list records across people, companies, deals, and custom objects, retrieve detail by ID, create, upsert, and update records, inspect attribute definitions, manage lists, comments, notes (with semantic search across note bodies), and tasks, and reach activity data including meetings, call recordings with transcripts, and emails, several with AI-powered semantic search, plus aggregate reporting. Close's server is built for high-velocity sales teams: search leads and contacts (including natural-language search), create and fetch leads, contacts, opportunities, notes, tasks, pipelines, and statuses, run activity searches and aggregations for reporting, and manage email and SMS templates, smart views, workflows, and voice agents. Both connect over OAuth and gate writes by scope. Here is how they compare.
How they compare
| Dimension | Attio | Close |
|---|---|---|
| CRM philosophy | Data-model-first and flexible: people, companies, deals, and fully custom objects, with attribute-definition introspection so the agent knows each object's shape. | Sales-velocity-first: a lead-centric model with contacts, opportunities, pipelines, and statuses tuned for outbound and inside-sales teams. |
| Activity and comms | Reaches meetings, call recordings with transcripts, and emails — several with AI-powered semantic search — plus notes with semantic search across bodies. | Manages email and SMS templates, smart views, workflows, and even voice agents, so an agent can log calls and draft and schedule follow-ups. |
| Search and reporting | Structured search and list tools across records and lists, semantic note search, and aggregate reporting via run-basic-report. | Natural-language lead search, activity search, pagination, and aggregations, plus reporting-metric discovery for pipeline questions. |
| Auth and write safety | Remote endpoint at mcp.attio.com/mcp over OAuth, scoped to the authorizing user and Attio's permission model; reads auto-approve while writes request confirmation in supported clients. | Remote endpoint at mcp.close.com/mcp over OAuth with Dynamic Client Registration; a Close-Scope header sets mcp.read, mcp.write_safe, or mcp.write_destructive to govern available tools. |
| Best-fit task | Teams on Attio that want an agent to work a flexible, custom-object CRM and mine activity data (calls, meetings, emails) with semantic search. | High-velocity sales teams on Close that want an agent to manage the pipeline, log activity, and drive outbound comms and workflows. |
Verdict
Both are modern CRMs an agent can fully operate over OAuth, so pick by team shape and where your data lives. Choose Attio if you want a flexible, data-model-first CRM with custom objects and rich activity mining — meetings, call transcripts, and emails with semantic search — and you like that writes request confirmation while reads auto-approve. Choose Close if you run a high-velocity sales team and want an agent to work a lead-centric pipeline, log activity, and drive email, SMS, and workflow automation, with a Close-Scope header that cleanly separates read, safe-write, and destructive-write access. Attio is the flexible, activity-rich CRM; Close is the sales-velocity machine. The decision follows whichever platform your revenue team already runs.
FAQ
- Can either CRM agent search by natural language?
- Both can. Close ships a natural-language lead search tool, and Attio offers AI-powered semantic search across notes and activity data like call transcripts and emails. The difference is emphasis: Attio leans on semantic search over activity, Close on fast lead and activity search for pipeline work.
- How do they control destructive writes?
- Close uses a Close-Scope header with mcp.read, mcp.write_safe, and mcp.write_destructive levels, so deletes require the destructive scope. Attio auto-approves reads while write operations request confirmation in supported clients, and access respects the authorizing user's permissions.