With the Teamdash MCP connector, you can use your Teamdash data inside AI assistants such as Claude and ChatGPT. Ask about your hiring statistics, search your talent pool or update candidates straight from a chat. Your Teamdash permissions and history tracking stay in place.
MCP (Model Context Protocol) is an open standard that lets AI assistants connect securely to other tools.
What you can do
Find and understand information
Search candidates and your talent pool, and open candidate profiles, including CVs, comments, messages, interviews and scorecards
View projects, tasks, activities, tags and users
Get statistics and indicators, and ask the AI to explain what a metric means
Take action
Add candidates to projects and move them between stages
Reject candidates
Add comments and tags
Create or update candidates
Create and complete tasks
Send emails to candidates
Make and edit landing pages
Find your landing pages and open any page to see its content, theme and settings
Edit page content in plain language: change texts, images and other blocks
Add new blocks, remove blocks you don’t need or change their order
Update page settings and the theme
Ask for a preview to see how the page or a single block looks
Publish the page when it’s ready
Every change is saved as a draft straight away, and you can watch it update live if the page is open in the editor. The page goes live only when you ask the assistant to publish it.
Example prompts
"How many candidates applied to our Sales Manager project last week, and which stages are they in?"
"Find candidates in our talent pool with warehouse experience in Tallinn."
"Draft a friendly follow-up email to the candidates in the Interview stage."
“Update the Sales Manager job ad: shorten the intro, add a benefits section and show me a preview.
Tip: Not sure where to start? Ask your assistant: "What can you do with Teamdash?"
How to connect
In Teamdash, go to Settings → MCP.
Follow the step-by-step instructions for your AI assistant (Claude or ChatGPT).
Log in to Teamdash and approve the connection.
Important: Only connect using the instructions in Settings → MCP. Before you approve, check that the request comes from the AI assistant you're actually using. Don't approve connection requests you didn't start yourself.
What the AI can see and do
The AI works with your user permissions. It sees and does exactly what you can see and do in Teamdash, no more and no less.
How actions are tracked
Everything done through the AI is saved in candidate history, under your name, with a AI Assistant badge for eg. "Via Claude" . Your team can always see what was done in Teamdash and what was done through an AI assistant.
Emails to candidates wait 5 minutes before they're sent. You can cancel them during that time.
Best practices
Check before you act. Always review rejections and emails to candidates before they go out. AI can make mistakes.
Treat candidate content with care. The AI reads text written by candidates, such as CVs and emails. If an answer or suggested action looks unexpected, check it in Teamdash.
Double-check important numbers against your Teamdash reports before sharing them.
For admins
Control who can connect
You can control who can connect to the MCP and stop connections via our Roles menu.
Go to role settings.
Remove the permission "Can connect AI assistants (MCP clients) to their own account".
This also cuts off that role's existing connections straight away.
To switch the connector off for your whole account, contact our support team.
Review connections
You can see and remove active connections in Settings → MCP.
Data protection
When you use the connector, the candidate data that the AI works with is shared with your AI provider (for example, Anthropic or OpenAI). That sharing falls under your organisation's own agreement with that provider, not under your agreement with Teamdash.
EU AI Act: The MCP is a data connector and contains no AI model. The AI system is the assistant you choose, and how you use it decides the risk class. Search, summaries and reporting are assisting tasks. Using it to rank, filter or reject applicants is a high-risk use under Annex III (employment), with deployer obligations on your company. We recommend a policy that humans make every hiring decision.
We recommend that you:
Use a business plan from your AI provider that includes a data processing agreement (DPA)
Make sure model training on your data is turned off
Mention your AI provider in your candidate privacy notice, if needed
Check with your data protection officer before rolling the connector out to your team
FAQ
Which AI assistants are supported?
Claude and ChatGPT.
What would be the use case of connecting TeamDash MCP to Claude and ChatGPT? How does this affect the EU AI Act or GDPR?
Use case: The Teamdash MCP lets a recruiter connect their own AI assistant (e.g. Claude) to Teamdash and work in plain language. For example: search the talent pool, summarise a candidate's profile and interview notes, ask "what changed in my projects / what are my tasks", pull hiring statistics, or move, tag and comment on candidates. It can also edit job ad pages. The assistant acts as the connected user and can only see and change what that user's role allows. Changes it makes show in candidate history marked "via Claude".
GDPR: When a user connects, the candidate data the assistant reads (contact details, CV text, comments, messages, interview transcripts, scorecards) goes to the AI provider under your company's own Claude/ChatGPT account. Your "Azure OpenAI (EU)" setting in Teamdash does not cover this, because this data does not go through Teamdash's AI vendors. Use a business plan with a DPA (no training on your data, EU transfer terms, retention) and add it to your records of processing.
EU AI Act: The MCP is a data connector and contains no AI model. The AI system is the assistant you choose, and how you use it decides the risk class. Search, summaries and reporting are assisting tasks. Using it to rank, filter or reject applicants is a high-risk use under Annex III (employment), with deployer obligations on your company. We recommend a policy that humans make every hiring decision.
Controls: We can switch MCP off for your whole instance. You can also allow or block it per role with "Can connect AI assistants (MCP clients) to their own account", and admins can see and revoke connections under Settings → MCP.
Can the AI see data I don't have access to?
No. It uses your Teamdash permissions.
Does Teamdash use my data to train AI?
No. Whether your AI provider uses your data depends on your plan and settings with that provider.
I sent an email by mistake. What can I do?
Emails wait 5 minutes before they're sent. Cancel the email in Teamdash during that time.
How do I disconnect?
Go to Settings → MCP and remove the connection.