Improve Fin resolution rate with session context (beta)

The Pendo integration with Intercom's Fin AI Agent gives Fin access to real-time session data from Pendo during active support conversations. Joint Pendo and Intercom customers can use this to help Fin understand what a visitor was doing in the app before contacting you, so it can deliver faster, more relevant answers.

Note: Pendo also offers a separate integration that lets Fin surface relevant Pendo guides directly in chat. The two integrations work well together, but each has its own setup procedure. If you want Fin to offer in-app walkthroughs from Pendo, see Serve Pendo guides using Intercom Fin.

Use cases

This integration lets Fin look up a visitor's recent session activity in Pendo when they start a support conversation. Pendo and Intercom customers can use this context to:

• Reduce back-and-forth by giving Fin visibility into what the visitor already tried.
• Improve resolution accuracy by grounding Fin's responses in actual user behavior, not just the visitor's description of the problem.
• Shorten time-to-resolution by skipping basic troubleshooting steps when session data shows the visitor has already completed them.

How it works

The integration connects Pendo session data to Intercom's Fin AI Agent so Fin can retrieve visitor context during a live conversation. The end-to-end flow works like this:

1. A visitor starts a support conversation in Intercom Messenger and describes an issue.
2. Fin calls the Pendo integration to retrieve the visitor's recent session activity, including Pages visited, tagged Features clicked, tagged Track Events called, and related frustration signals. Untagged clicks and Track Events are omitted from the returned activity timeline.
3. With the session data available, Fin can reference what the visitor was doing, acknowledge steps already taken, and provide a more targeted answer.
4. The session data lookup is configured through a data connector in Intercom. Pendo surfaces the session data through the MCP server, and Fin references it as part of its response generation.

Tip: For best results, give your Pages and Features descriptive names and descriptions in Pendo. Fin uses these names when referencing session activity, so clear labels like "Billing Settings Page" or "Export CSV Button" help Fin communicate precisely about where the visitor was and what they did.

Before you begin

To complete this setup, you need:

• An Intercom workspace with Fin enabled.
• An agent set up in Agent Toolkit with the MCP connection in place. This creates the authentication token and Pendo MCP data connector in Intercom that the session context tool relies on. See Set up an agent in Agent Toolkit.
• Semantic search turned on in Pendo Subscription settings > AI access. If this setting is off, your subscription may have opted out of Google AI features, and the session context integration with Fin is unavailable.

Step 1. Set up your agent and connect MCP

The agent, authentication token, and Pendo MCP data connector are all set up in the Agent Toolkit setup wizard. If you haven't done this yet, follow Set up an agent in Agent Toolkit, then return here to enable the session context tool in Fin.

Step 2. Enable the GetSessionActivity data connector in Fin

The GetSessionActivity data connector exposes the GetSessionActivity tool to Fin. This is the tool Fin uses to look up a visitor's recent Pendo session activity and base its answers on what the visitor actually did in your app.

In the Intercom app:

1. Go to Settings in the bottom-left corner.
2. Under Integrations, select Data connectors. The Pendo MCP server you added during agent setup is listed.
   
   
   
3. Find the MCP server you added during agent setup, then select New Pendo Support Agent MCP connector.
   
   
   
4. Select the GetSessionActivity tool, then select Add connectors. This populates the data connector in the table.
5. Select the newly added GetSessionActivity from the table under the Name column, then select Edit in the top-right corner.
   
   
   
6. Under Data inputs, select + Data input.
7. For Data source, select Select data attribute, then choose the Intercom attribute that uses the same value as the Pendo Visitor ID so Fin can pass the user ID into the connector. Use this same value in the remaining steps.
   
   
   
8. Select Save.
9. Scroll down to the Body section. With your cursor, click between the two double quotes here: "arguments":{"visitorId":"", then select Add data and choose the same attribute you added in step 7. This maps the selected Intercom attribute to the visitorId argument in the GetSessionActivity tool.
   
   
   
10. Scroll down to the Test section, and enter a Test value that matches the format of the Visitor ID in Pendo.
11. Optionally, we recommend allowing Fin to use this connector outside of a workflow or procedure. To do this, open the Fin tab, then turn on the Directly toggle.
    
    
    
12. When ready, select Set live in the top-right corner to make the data connector available in Fin.

Once the data connector is live, Fin can call the GetSessionActivity tool whenever a visitor starts a conversation. Fin queries Pendo for recent session context and uses this information to tailor its responses during support interactions.