New research: 1,000 enterprise leaders on the AI decision-confidence gap.

Download the report →

Agent MCP connectors

This article is for product managers, researchers and experience teams who want the TheyDo Agent to work across the tools their team already lives in, not just the data already in TheyDo. MCP Agent connectors let the Agent read from and write to tools such as Jira, Dovetail, Amplitude and Miro, so experience context flows both ways.

Note: MCP Agent connectors are currently in private beta. To take part, ask your CSM about access.

What connectors are

MCP Agent connectors are built on MCP (Model Context Protocol), an open standard that lets AI agents connect securely to other software. Many of the tools your team uses now offer an MCP server, and connectors plug the TheyDo Agent into them. You don't need to know anything about MCP to use them: you connect a tool once, then ask the Agent in plain language.

Your customer context is spread across many tools: interview recordings in a research repository, delivery work in Jira, usage data in an analytics tool, journey maps in Miro. Instead of exporting a file and importing it into TheyDo, you ask the Agent. It reads from the source, turns what it finds into TheyDo building blocks such as insights or solutions, and places them in the right journey. Where it makes sense, it links back to the original item so anyone can trace an insight to its source.

Connectors work in both directions. Besides pulling context in, the Agent can push context back out, for example adding customer evidence to a Jira issue or posting an update to Slack, as long as your account in that tool allows it.

Why use connectors

  • Bring every experience signal into one place. Capture the context that lives in the tools you already use, instead of leaving it stranded there.
  • Feed your journey graph. Let the Agent process raw research, notes and delivery work and map it to the right parts of the customer experience.
  • Keep other teams in the loop. Send validated customer evidence to the tools where delivery and product teams already work.

Tip: Connectors bring context into TheyDo through the Agent. If you'd rather ask questions about your TheyDo data from an outside AI tool such as Claude, see TheyDo MCP (read-only).

How permissions work

Connectors never give the Agent more access than you have.

  • In TheyDo, the Agent works within your normal workspace permissions.
  • In the connected tool, the Agent sees and does only what your own account in that tool is allowed to. If you can't open a Jira project, the Agent can't either.
  • Each connection is personal. Connecting Jira gives the Agent access to your Jira account, not your colleague's. Each person who wants to use a connector connects it themselves.
  • Some tools need an admin on your side to allow the connection first. How to set up a TheyDo Agent MCP connector lists which ones.

How to use connectors with the Agent

  1. Open the TheyDo Agent.
  2. Connect the tool you want to use: click the + icon, select Manage connectors, open the tool and click + Instance. See How to set up a TheyDo Agent MCP connector for what each tool needs.
  3. If you want the Agent to create or change anything in TheyDo, turn on Edit mode. To only read and summarize, you can leave it off.
  4. Ask in plain language and name the tool, for example: "Pull the findings from our latest Dovetail onboarding study into the Onboarding journey as insights."
  5. Review what the Agent proposes and approve the changes you want to keep.

When you ask the Agent to use a connector, or use a skill that relies on one, it tells you the connector's status in the conversation, for example whether it's connected or needs reauthorizing. That way you know straight away if something needs fixing before it starts.

Writing back to an external tool, such as commenting on a Jira issue, follows that connector's settings. If its tools are set to ask for approval, the Agent pauses and asks you before it acts in the other tool.

Tip: Name the tool and the destination in your request. "Bring the in-progress Jira issues for the Checkout epic into the Checkout journey as solutions" works better than "get my Jira stuff".

What you can do with connectors

Here are the main ways teams use connectors today. Which MCP connectors does the TheyDo Agent connect to? shows which tools support each one.

Pull research and transcripts into TheyDo

Bring sessions and findings from research tools such as Dovetail, Condens, Marvin or Lookback into TheyDo. The Agent extracts pain points, needs and verbatims, creates them as insights tagged with their source, and places them on the relevant journey steps.

Try: "Import the key findings from our Dovetail project 'Q3 onboarding interviews' as insights on the Onboarding journey."

Keep journeys aligned with delivery work

Bring in-progress Jira or Linear issues into TheyDo as solutions, linked to the journey stages they improve. You can also pull a spec from Confluence or Notion and create a solution from it, so delivery context sits next to the evidence behind it.

Try: "Create solutions on the Checkout journey for the in-progress issues in our CHK Jira project, and link each one back to its issue."

Bring customer evidence to delivery teams

Send a validated insight to the Jira issue it should inform, so engineers see the customer evidence without leaving Jira. This needs permission to edit issues in Jira.

Try: "Add the three highest-scoring insights about payment errors as a comment on PAY-142, with links back to TheyDo."

Align opportunities with your product roadmap

Link the opportunities in your journeys to the matching items on your Productboard roadmap, or push a validated opportunity into Productboard so it can be prioritized alongside the rest of the roadmap.

Try: "Find the Productboard features that address the top three opportunities on the Checkout journey and link them."

Back up journeys with product analytics

Query Amplitude, Mixpanel or PostHog for usage trends or funnel data and add the result as evidence on a journey step or opportunity, with a link back to the chart. Session-replay tools such as Fullstory, LogRocket and Contentsquare work the same way for friction on digital steps.

Try: "How many users dropped off between account creation and first project last month? Add the finding to the Activation step."

Create or enrich a journey from a Miro journey map

Point the Agent at a journey map in Miro and it can either create the matching journey in TheyDo, with phases and steps filled in, or enrich a journey you already have with what's on the map.

Try: "Create a journey from the 'Claims journey map' board in Miro." or "Add the pain points from the 'Onboarding workshop' journey map in Miro to the Onboarding journey."

Note: The Miro connector is for journey maps only, not for other kinds of boards. Sticky notes and content outside the main journey structure may not all come across in one go. Check the result and ask the Agent to add anything missing.

Search team conversations

Search Slack for context while you research a journey step, or have the Agent post an update, such as a new insight, to a channel.

Connectors and integrations

TheyDo also offers full integrations, for example with Jira, Qualtrics and Snowflake, which an admin sets up once for the whole organization and which sync data on a schedule. Connectors are lighter: each person connects their own account and the Agent pulls context when you ask. If your organization needs centrally governed, always-on data flows, an integration is usually the better fit. What are integrations in TheyDo? explains the difference in more detail.

Further reading