Skip to content
How we review
ai-agent.review — AI Agent Reviews, home
← All news

Product updates

Intercom Fin flags chats that may have started with a customer’s AI agent.

Vendor-reportedChecked: [release][docs]

Editorial attribution: Gosia Madej

Prepared with AI assistance. Editorial attribution does not imply personal product testing or manual review of every story.

Published: · Event date:

Intercom now labels new conversations in its web Messenger to help support teams see when a personal AI agent may be speaking for a customer.

Vendor-reportedChecked: [release][docs]

What changes for users

Start with an Inbox view rather than a different level of service. Review samples before routing these conversations differently, and interpret the report as a view of predicted labels, not a count of verified agent traffic.

Editorial assessmentChecked: [release][docs]

Two frosted speech bubbles face each other; one has a small amber tag.
AI-generated editorial illustration. Not a product screenshot. · AI Agent Reviews (AI-generated editorial illustration)

The November 13 change adds a read-only rating: Yes, Likely or No. Intercom says it is available on all plans and enabled by default. The rating is set before Fin’s first reply and does not cover older conversations, mobile SDKs, email or voice.

Vendor-reportedChecked: [release][docs]

Teams can filter their Inbox and compare resolution and escalation rates in a dedicated report. Fin’s replies do not change by default; operators can use the rating in routing and procedures they configure.

Vendor-reportedChecked: [release][docs]

Checked against sources

AI-assisted source check

How this was checked

We compared the announcement with Intercom’s setup and limitations documentation.

Editorial assessmentChecked: [release][docs]

We did not test detection accuracy. Intercom says the rating can misclassify both people and agents.

Editorial assessmentChecked: [release][docs]

Source evidence

Original source excerpt. This shows what the source says; it does not establish product performance.

The Personal agent involved rating is an estimate, not proof.

Read original source [docs]

Sources