Attio MCP: connect Attio to Claude and ChatGPT
Ask Claude or ChatGPT which deals in your sales pipeline are still in Negotiation, what the team last wrote about Fjällström Logistics or which tasks you have due this week. The answer comes straight from Attio, with a link to each record.
With Deployed OS everyone connects with their own Attio login. The AI sees the records and lists that person can see, and the only things it writes are notes and the person’s own tasks, when they ask.
8 read actions · 2 write actions · Works in Claude, ChatGPT and Copilot
What your team can do with Attio in Claude and ChatGPT
Know the account before the call
The pipeline in plain words
Log the call, set the follow-up
Any list, not only sales
Ask it like this
“What do we have on Fjällström Logistics: people, open deals and the latest notes?”
“Which deals in the Sales pipeline are in Negotiation, and who owns each one?”
“Add a note to Nordhavn Energy: met Erik, they want a pilot for two teams starting in November.”
“Create a task for me to send the revised proposal to Kestrel Analytics on Friday.”
“What tasks do I have open this week, soonest first?”
“List the companies in our Partnerships list that are still at Intro call.”
What the Attio MCP connector does today
- Reads
attio_searchSearch the people, companies and deals you can see by words in their name, email or domain, up to 25 at a time. It uses Attio’s fuzzy search, which is in beta.
- Reads
attio_list_recordsList the records of one object you can see, such as people, companies, deals or one of your own objects, filtered and sorted the way you ask. Up to 50 at a time, with every value and a link.
- Reads
attio_get_recordRead one record: every value, its link, its newest notes (up to 10), its open tasks soonest due first, the lists it is in and its latest comments.
- Reads
attio_list_objectsList the workspace’s objects: people, companies, deals when they are turned on, and your own custom objects.
- Reads
attio_list_attributesList the attributes of one object or list, with the stages of its status attributes, such as a deal’s stage, so the AI can filter on them.
- Reads
attio_list_listsList the lists you can see, such as pipelines: name, which objects they hold and the access the workspace has.
- Reads
attio_list_entriesList the entries of one list, optionally filtered on the list’s own attributes such as its stage, each with the record it holds by name and link. Up to 25 at a time.
- Reads
attio_list_tasksList your own tasks: the open ones, soonest due first, or the completed ones, latest first, each with the records it is about.
- Writes
attio_create_noteAdd a plain-text note with a title to one record, as you, once you have confirmed the text. Nobody is mentioned or notified, and nothing else on the record changes.
- Writes
attio_create_taskCreate a task assigned to you, never to anyone else, optionally about one record and with a due date, once you have confirmed the text. Nobody is notified.
Need it to do more? We extend connectors for the companies that use them.
Who sees what
- Reads, plus notes and your tasks
- The connector reads people, companies, deals, your own objects, lists, notes, tasks and comments. It can add a note and create a task for you. It cannot edit values, move an entry to another stage or delete anything.
- Each person’s own rights
- Everyone connects their own Attio login. Records and lists a person can’t open in Attio stay hidden from the AI as well.
- Writes only with Edit access
- Admins choose which groups get Attio, and only groups with Edit access can add notes and tasks. Each one is written as the person asking, after they confirm the text, and tasks are only ever assigned to them.
- Every call logged
- Who asked, which connector, when. Deployed OS shows reads and writes per connector and per person, and the log can be exported.
Set up the Attio connector in three steps
Attio’s own MCP server or Deployed OS?
Attio runs its own hosted MCP server. You sign in with your Attio account and it works with the same permissions as your Attio user. It can search, read, create and update records, lists, notes and tasks, and search meetings, call recordings and emails. For one person who wants Attio in Claude or ChatGPT, it is a good fit.
Deployed OS is for when the whole company should use it. One company connection for Attio and all your other tools, with each person on their own login, admins choosing which groups get Attio and which of them may write, and a log of every call in one place. The same connection works in Claude, ChatGPT and Copilot, next to your company’s skills and guardrails.
Attio MCP: common questions
Yes. Attio has its own MCP server, and it suits one person well. Deployed OS gives the whole company one connection for Attio and your other tools, with each person’s own login, access per group and a log of every call.
Two things: it can add a plain-text note to a record and create a task assigned to you, when you ask and confirm the text. Nobody is mentioned or notified, though your workspace’s own Attio workflows may still run. No value is edited, no entry changes stage and nothing is deleted. If you want it to do more, we can extend it for you.
Your admin sets up Attio in Deployed OS once. You add your company’s Deployed OS connection in Claude, sign in to Attio with your own login and start asking. The same connection works in ChatGPT and Copilot.
Only if they can in Attio. The AI uses each person’s own Attio login, so the access you have set on objects and lists in Attio applies as it is.
Yes. The AI looks up your workspace’s objects, attributes and list stages before it filters, so a custom object such as Investors or Candidates works the same way as people and companies.
Deployed is an EU company and fully GDPR compliant, and you choose where your data is stored, down to the city. Attio data is fetched when someone asks for it.
It depends on the size of your team. Get in touch and we will walk you through it.
See it on your own tools.
Tell us which systems your company runs on. We’ll show you what your team can ask, and what IT gets to control.
