One really simple n8n workflow supercharged our AI support agent's resolution rate. Here's the breakdown:
Background
From the RB2B app we send events to Intercom whenever a user does a thing — purchases, changes plans, turns something on/off, starts a trial, ends a trial, has an integration error — you name it.
The Hiccup
That's all well and good for our human team. We have eyes on that. But Fin doesn't.
I don't know why it doesn't, but that's okay.
Since it couldn't see that information natively, I decided to give it that information via a data connection.
How It Works
Every time a user connects with our support agent, we feed the agent a list of recent events, giving it more context into recent activity on the user's account.
- Send the user's email address to n8n
- Hit Intercom's "Events" API and pull a list of recent events
- Split the array into individual events
- Filter out anything older than 60 days and anything irrelevant (recent article views, etc.)
- Update the Unix timestamp to something the user can read (e.g. September 2, 2025 12:34PM ET)
- Send it back to Fin to use for the ticket
Results
Our resolution rate has jumped by more than 7% since implementing this a couple weeks ago.
Bonkers.
