Customer support · Production

Rocket Money operationalizes AI with Intercom Fin, resolving 68% of support conversations autonomously

The problem

As Rocket Money scaled to over 60,000 monthly support conversations, button-based routing workflows placed the burden of precision on customers; misclicks sent conversations to an unassigned inbox requiring manual rerouting, and one teammate was spending two to three hours a day on this work alone.

First attempt

Button-based routing workflows were thoughtfully built but could not anticipate every support scenario, creating an unassigned inbox backlog and making manual triage unavoidable at scale.

Workflow diagram · grounded in source
1
Inbound support query routed to Fin
trigger
“they expanded to key workflows that could be clearly scoped and carefully controlled, like billing management, app troubleshooting, and account access requests”
2
Fin handles initial interaction
ai_action
“While Fin handles the initial interaction and many common queries”
3
Routing rules gate queries
routing
“Routing rules and logic paths were built to ensure Fin only handled queries it was equipped to resolve, and that high-risk or sensitive issues were escalated immediately”
4
Fin resolves conversation
output
“resolves 68% of them, which amounts to tens of thousands each month”
5
Human agents handle escalations
human_review
“They step in when oversight or escalation is the best path forward”
6
Agents optimize Fin continuously
feedback_loop
“Human agents who once spent hours manually rerouting conversations, now focus on optimizing Fin: refining workflows, identifying edge cases, and improving how Fin handles exceptions”
Reported outcome

Fin now handles over half of all conversations and resolves 68% of them; manual triage has been eliminated, human CSAT has risen by six points, and the team unlocked nearly $1M in annual efficiency gains.

Reported metrics
Fin conversation resolution rate68%
conversations involving Finover half of all conversations
human CSAT increasesix points
Annual efficiency gainsnearly $1M
Show all 7 reported metrics
Fin conversation resolution rate68%
conversations involving Finover half of all conversations
human CSAT increasesix points
annual efficiency gainsnearly $1M
email billing management CSAT80%+
monthly support conversation volumeover 60,000 conversations each month
manual rerouting time per agent (before)two to three hours a day
Reported stack
FinIntercom
Source
https://www.intercom.com/customers/rocket-money
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Fin now handles over half of all conversations and resolves 68% of them; manual triage has been eliminated, human CSAT has risen by six points, and the team unlocked nearly $1M in annual efficiency gains.

What tools did this team use?

Fin, Intercom.

What results were reported?

Fin conversation resolution rate: 68%; conversations involving Fin: over half of all conversations; human CSAT increase: six points; Annual efficiency gains: nearly $1M (source-reported, not independently verified).

What failed first in this deployment?

Button-based routing workflows were thoughtfully built but could not anticipate every support scenario, creating an unassigned inbox backlog and making manual triage unavoidable at scale.

How is this customer support AI workflow structured?

Inbound support query routed to Fin → Fin handles initial interaction → Routing rules gate queries → Fin resolves conversation → Human agents handle escalations → Agents optimize Fin continuously.

Related customer support cases