AI customer service on HubSpot conversations with human escalation
Challenge
Recurring requests that saturate support, without an automatic first filter.
Solution
A flow that classifies requests and prepares AI draft replies, with human escalation on complex cases.
Systems involved
- HubSpot
- OpenAI
- Make.com
Before
The team answered many similar requests by hand, with long response times and little priority for complex cases.
After
Requests are classified and receive an AI draft, while complex cases pass to a human operator.
Result
Less manual work on recurring requests and clearer priority for complex cases.
A customer service team received many similar requests on HubSpot conversations. The team answered by hand, case by case, even when the question was recurring. This lengthened response times and left little room for the cases that truly needed attention.
We built a Make.com scenario that classifies each request with OpenAI and, for known cases, prepares a draft reply in the HubSpot thread. The operator confirms or edits the draft before sending, so the control stays human. Complex or ambiguous requests do not receive drafts. They are flagged for escalation to a person, with the right priority.