🔐 Who can use this feature?
Roles: All users, once enabled for the account. An account administrator registers the organization at cultureamp.com/mcp and turns the connection on before anyone can connect.
Subscriptions: Any
MCP (Model Context Protocol) is an open standard that lets AI assistants like Claude and ChatGPT connect directly to the apps and data you use, including Culture Amp.
Availability
Claude and ChatGPT are the AI tools Culture Amp has tested today, and we're extending support to other AI tools over time.
If your organization signs in to Culture Amp through an older SAML single sign-on setup, that method doesn't work with this connection yet, so everyone connecting will need another sign-in option, such as Google or a Culture Amp password.
To get started, an Account Administrator needs to complete self-serve registration. Once Culture Amp confirms your organization's eligibility, your team can connect.
What Culture Amp MCP does
The connection uses your own Culture Amp sign-in, so the AI assistant only sees what you can already see in Culture Amp. Once it's connected, Claude or ChatGPT can reach three areas of your account:
Your engagement survey results: available if your organization uses Engage.
Your performance feedback: available if your organization runs performance reviews in Culture Amp.
Your 1-on-1 conversations: find them and work with their agenda topics, action items, and notes, all without opening Culture Amp yourself.
Each area is covered in full further down, under Available tools.
How it works
Setting this up has two sides: an account administrator turns the connection on once for the whole organization, and then each person who wants to use it connects individually.
Who | What they do |
Account administrator | Registers the organization at cultureamp.com/mcp, waits for Culture Amp to confirm the account is enabled, switches Activate MCP on, and shares the resulting connection address with their organization. |
Each user | Adds Culture Amp as a custom connector in Claude or ChatGPT using that address, signs in with one of their organization's existing Culture Amp sign-in options, and approves access. |
Once an administrator has turned the connection on, each person's own connection sequence looks like this:
Add Culture Amp as a custom connector in Claude or ChatGPT, using the connection address your administrator shared.
Sign in on the Culture Amp sign-in screen that appears.
Review the authorization screen and approve access.
Start asking the assistant about your survey results, your performance feedback, or your 1-on-1s.
The full step-by-step instructions for both sides, including what to do if something doesn't work the first time, are under Setting things up below.
Benefits of connecting
Work where you already are: Ask about your survey results, performance feedback, or 1-on-1s from inside Claude or ChatGPT, instead of switching over to Culture Amp to look things up.
Nothing new to manage: The connection uses the same Culture Amp sign-in and permissions you already have, so there's no separate account or password to set up.
One connection, three areas: A single setup covers your engagement survey results, performance feedback, and 1-on-1 conversations, instead of three separate integrations.
Always current: Because the assistant reads your Culture Amp data directly, what it tells you is as current as Culture Amp itself, not a stale export.
Available tools
Your engagement survey results
If your organization uses Engage, you can also ask Claude or ChatGPT to:
Find your surveys: find your engagement, pulse, and check-in surveys.
See participation and scores: for the whole company or narrowed to a group such as a location, department, or your own team.
Ask what's driving a result: and where the survey's focus areas suggest paying attention.
See question-level answers: see how people answered a multiple-choice question.
Compare scores: compare a score against the comparisons that survey's report already offers, such as a benchmark, or see how it has changed since the last survey.
You see the same surveys and results you can see in Reports: if you're an account administrator, that's every survey that's launched, and otherwise it's the surveys whose reports have been shared with you.
💬 Try asking: "Summarize how engagement moved since the last survey, which factors moved most, and how participation compares for the [role name] in [location]"
Your performance feedback
If your organization runs performance reviews in Culture Amp, you can also ask Claude or ChatGPT to:
Pull together feedback and context: gather feedback and profile information to help you prepare for a conversation or review, such as peer feedback, upward feedback, self-reflection, past manager reviews, and anytime feedback, for yourself or for one of your direct reports.
Find the right person: find a specific person by name or email, or list your direct reports, so you can point it at the right person.
The assistant can only pull together past feedback and profile context like this: it can't open or show you a review you're currently writing, or its questions and answers, so you'll still write and submit each review yourself in Culture Amp. It can also only reach your own feedback and profile information, or that of people who report directly to you.
💬 Try asking: "Pull together the feedback [employee name] has received this year to help me prepare for her review."
Your 1-on-1 conversations
Once you've connected Claude or ChatGPT to your Culture Amp account, you can ask it to:
Find your conversations: find or list your 1-on-1 conversations, optionally narrowed to a person's name or to a relationship such as your manager, a direct report, or a colleague.
View an agenda: show the agenda for a specific 1-on-1, including open topics, open action items, and anything completed in about the last month.
Add a topic: add a new agenda topic, with an optional note giving more detail.
Add an action item: add one, and assign it to either participant in the conversation.
Add a note: add a note to a topic.
Mark a topic discussed: mark an agenda topic as discussed, or mark it as not discussed if you did that by mistake.
Mark an action item complete: mark it as complete, or mark it as incomplete if you did that by mistake.
Anything you add or complete this way shows up the same as if you'd made the change directly in Culture Amp. For example, a note the assistant adds appears on that conversation's agenda just like a note you typed in yourself. See 1-on-1 Conversations for how conversations, agendas, and notes normally work in Culture Amp.
💬 Try asking: "What's on my 1:1 agenda with [employee name], and can you add a topic?"
Setting things up
Register and turn on the connection
An account administrator needs to register, then turn the connection on for your organization, before anyone can connect.
As an account administrator, register your organization at cultureamp.com/mcp.
Wait for Culture Amp to review your account and confirm whether and when the connection is set up for you.
This isn't automatic or instant, so allow some time before you hear back.
We assess each registration for fit, so registering doesn't guarantee access.
Once Culture Amp confirms your account is enabled, go to Administration > Account > MCP and switch Activate MCP on.
This is how your organization opts in, so it needs to be done by an account administrator before anyone signs in.
An account administrator can switch it back off later.
Copy your organization's connection address from the MCP link field, which appears on that same screen once Activate MCP is on.
Share that address with the people in your organization who want to connect. Each person needs it to complete the next section.
Connect your AI assistant
Every person who wants to use the connection does this once, for themselves. Your organization has its own connection address, so there's nothing to pick from a list of available apps: you add Culture Amp as a custom connector using that address.
In Claude or ChatGPT, find the option to add a custom connector, then choose to enter a connector address.
Each AI tool uses its own wording and location for this, so check that tool's own help if you can't find it.
Enter your organization's connection address exactly as your administrator shared it with you.
The address ends with your organization's own Culture Amp name. Leaving that last part off is the most common reason the connection fails, so paste the whole address rather than typing it.
Sign in on the Culture Amp sign-in screen that appears.
Use one of the sign-in options your organization has already set up for Culture Amp, such as Google or your Culture Amp email and password.
Review the authorization screen and approve access.
You're letting the AI tool see your survey results and performance feedback, and see and act on your 1-on-1s, on your behalf, matching what you can already see and do yourself in Culture Amp, nothing more. The exact wording on the screen can differ depending on how your organization was set up.
You might be asked to sign in or approve access again at some point later on.
Start asking the assistant about your survey results, your performance feedback, or your 1-on-1s.
If Culture Amp doesn't appear straight away, fully quit the AI tool and reopen it.
Supported AI tools and sign-in options
The connection is built to the open MCP standard, so it isn't tied to one brand of AI assistant. A tool can only finish signing in if its own sign-in return address is registered with Culture Amp, so a tool we haven't tested might not be able to sign in, even once we extend support to it.
If your organization signs in to Culture Amp through an older SAML single sign-on setup, use Google or a Culture Amp password instead when connecting, since SAML doesn't work for this connection yet.
What the assistant can't reach
Your engagement survey results
Your engagement survey results
It only covers engagement, pulse, and check-in surveys: Draft surveys, and onboarding, exit, and 360 surveys aren't available through this connection yet.
It's read-only: The assistant can see survey results, but it can't change a report's filters, sharing, or anything else about a survey.
It only sees what you already see: Results are limited to the surveys and reports already shared with you, and the same confidentiality rules apply as everywhere else in Culture Amp. A result with too few responses to protect anonymity is withheld, not shown with a smaller group.
Some shared reports can't be read here: a report shared with you as a summary or participation view only opens in Culture Amp.
Your performance feedback
Your performance feedback
It can retrieve past feedback and profile context, but not a review you're currently writing: The assistant can pull together an employee's self-reflection, peer feedback, upward feedback, past manager reviews, and anytime feedback, but it can't list, open, or show the questions and answers on a review that's in progress.
It's read-only: The assistant can retrieve this feedback and context to help you prepare, but it can't write or save anything back. You always write and submit a review yourself in Culture Amp.
It only reaches your own feedback history, or your direct reports': You can ask about feedback and profile context for yourself or for people who report directly to you. It can't reach anyone else's performance information, even if you can find that person by name.
Your 1-on-1 conversations
Your 1-on-1 conversations
It covers your 1-on-1s: The assistant can only work with your 1-on-1 conversations, meaning the conversation itself, agenda topics, action items, and notes. Beyond your 1-on-1s, it can also reach your engagement survey results and your performance feedback, both covered earlier in this section, and nothing else in Culture Amp.
The assistant can add notes, but it can't read notes back: When you ask for an agenda, the assistant sees the topics and action items on it, including which ones are done, but not the text of notes already saved on them. So you can ask it to add a note, and you can read your notes in Culture Amp as usual, but asking the assistant to summarize or pull up your existing notes won't work.
The assistant only sees what you already see: It can only see and act on 1-on-1s you're personally a participant in, whichever side of the conversation you're on. This matches exactly what you see when you open Culture Amp yourself. It can't see a 1-on-1 you aren't part of, such as a direct report's 1-on-1 with a different manager.
How this differs from the Culture Amp API, Personal Notes, and Assigned Topics
This isn't the Culture Amp API
This isn't the Culture Amp API
Culture Amp also offers a separate Culture Amp API used for reporting and other integrations, set up with a developer account and API credentials. This connection is different: it uses your normal Culture Amp sign-in, not a developer account, API key, or separate credential setup.
Notes added this way aren't Personal Notes
Notes added this way aren't Personal Notes
Notes added this way aren't Personal Notes: When the assistant adds a note, it adds the shared note that both people in the 1-on-1 can see, the kind described in 1-on-1 Conversations. It isn't the separate, more private Personal Notes feature described in that same article, and the assistant can't create, read, or touch Personal Notes at all.
Adding a topic this way isn't an Assigned Topic
Adding a topic this way isn't an Assigned Topic
When you ask the assistant to add a topic, it adds an ordinary agenda topic, the same as if you'd typed it in yourself. This is different from Assigned Topics, which is a separate, admin-managed feature that schedules topics based on demographics. The assistant can't create, schedule, or manage Assigned Topics. If one is already on your agenda, it shows up like any other topic and can be marked as discussed.
💡Limitations
Your Culture Amp data is shared with the AI tool you connect: Using this connection means whatever you ask about (survey results, performance feedback, or 1-on-1s) is shared with whichever AI tool you connect. If you have questions about how that data is handled once it reaches the assistant, refer to that provider's privacy policy or contact the provider directly.
Very heavy use can be slowed down: Limits are in place to keep the service reliable for everyone, so a burst of many requests in a few seconds can be temporarily slowed. Normal day-to-day use won't reach them.
You'll be asked to sign in again from time to time: A connection you haven't used for a few days, or one you've had for about a month, will ask you to sign in again. Sign in again and the connection carries on where it left off.
Troubleshooting
The connection fails, or returns a "not found" error: Copy the address again from the MCP settings screen and paste it in full. The address has to end with your organization's own Culture Amp name, and leaving that last part off is the most common cause.
Culture Amp doesn't appear after you add it: Fully quit the AI tool and reopen it. If you'd added the connection before and it stopped working, remove it and add it again.
Sign-in fails: Use the sign-in option your organization has set up for Culture Amp. If your organization uses an older SAML single sign-on setup, use Google or your Culture Amp password instead, because that method doesn't work for this connection yet.
You sign in, but the assistant says the tools aren't available for your account: Ask an account administrator to open Administration > Account > MCP and check that Activate MCP is on. If it's already on, open a Support Conversation.
A survey or result doesn't show up, or says it isn't accessible: This matches what's shared with you in Reports. If you believe you should have access, check with whoever manages sharing for that survey, or open a Support Conversation.
Someone's performance feedback doesn't show up, or says it isn't accessible: This only works for your own feedback and profile information, or for people who report directly to you. If you believe it should be accessible and isn't, open a Support Conversation.
Something you asked the assistant to do doesn't appear in Culture Amp: Open that conversation in Culture Amp and check. If it didn't save, or doesn't match what you asked for, open a Support Conversation.
FAQ
Who do I speak to at Culture Amp to get access?
Who do I speak to at Culture Amp to get access?
There's no need to reach out to a Culture Amp contact directly. Register your organization at cultureamp.com/mcp and someone will be in touch after that's submitted.
Does this cost anything extra?
Does this cost anything extra?
No, there's no additional cost for the connection today.
How do I stop the AI tool from reaching my Culture Amp data?
How do I stop the AI tool from reaching my Culture Amp data?
The most effective way is for an account administrator to switch Activate MCP off in Administration > Account > MCP, which stops the connection working for everyone in your organization.
Removing the Culture Amp connector in the AI tool's own settings stops you using Culture Amp from that tool, but it doesn't end the sign-in you already approved, which stays valid until it expires. If you need access cut off straight away, use the administrator option above.
➡️ Next steps
Register for Culture Amp MCP: Sign up your account for access.
Guide to report sharing: See how survey report access and sharing work directly in Culture Amp.
Manager's Guide to Performance Reviews: See how to manage, complete, and share performance reviews directly in Culture Amp.
1-on-1 Conversations: See how conversations, agendas, and notes work directly in Culture Amp.
💬 Need help? Open a Support Conversation to speak with our Product Support team.