01 Company claim
The documented mechanism
Intercom's help center describes Fin as an AI agent that answers customer questions across chat, email, phone and other channels. Admins configure how and when Fin triages or hands off to human support. In Workflows, Fin escalates on natural-language cues: a direct request for a human, negative feedback, or repeated replies without resolution. Admins can branch the experience by audience, for example handing paying customers to a teammate while pointing non-paying customers to the Help Center. They can also have Fin ask for more details before handover. Intercom says Fin always hands off automatically when that is the safest option, including for high-risk content.
02 Kept apart from the practice
Reported outcome
Self-reported by Intercom (vendor): Intercom says it tested the 'ask for more information before handover' feature extensively and saw an increase in answer rate, confirmed resolutions and CSAT. No figures are given.
03 Read before copying
Confounders and limits
This is vendor documentation of a commercial product, and the page also calls Fin 'the best-performing AI agent for customer experience' without evidence. It describes configurable capabilities, not how any particular company's support performed. Results depend heavily on the quality of the knowledge content Fin learns from. Intercom's internal test conditions aren't described.
04 Over time
Known revisions, reversals or failures
No documented revision or reversal found in the sources reviewed. The workflows article documents behavioural limits: Fin doesn't respond to quick-reply button selections, and when a customer doesn't reply to Fin's request for more information, the conversation escalates and is not counted as a resolution.
05 Editorial interpretation — not a company claim
WeCaseStudy interpretation
WeCaseStudy interpretation: the documented design pattern is AI first, with explicit escalation triggers, audience-based routing, and gathering context before handover so human agents start with more information. For ai_triage_escalation, the decision a small company must make is really the escalation rules and which segments always get a human. Choosing the tool is the smaller question.
06 What transfers
Adaptation boundaries
A small company should not assume vendor-reported improvements will carry over. They depend on having maintained help content, admin capacity to configure and test workflows, and staffed human coverage for escalations (Fin respects workspace office hours on handover). The page says Fin is available on paid plans. Pricing and plan details weren't reviewed here and change over time.
Claims and their sources
Each material claim has its own type, source, location and review state. A documented policy is not proof that everyone follows it; a vendor describing its own product is vendor-reported.
| Claim | Type | Source and location | Review |
|---|
| CLM-0030 Intercom states that Fin's handoff can be configured, and that Fin always hands off automatically when that is the safest option for the customer, including when it detects high-risk content. | Documented practice | Fin AI Agent explained (Vendor documentation)
Section 'Human handoff': "Fin will always automatically hand off when that is the safest option" | Source checked — source checked 2026-10-17 |
| CLM-0031 Intercom describes Fin's model layer as including a model that detects when to escalate to humans. | Documented practice | Fin AI Agent explained (Vendor documentation)
Section '3. The Model Layer': "A model that detects when to escalate to humans." | Source checked — source checked 2026-10-17 |
| CLM-0032 Intercom documents that in Workflows Fin escalates based on natural-language cues such as a direct request for human support, negative feedback, or repeated replies without resolution. | Documented practice | Use Fin AI Agent in Workflows (Vendor documentation)
Section 'Configure Fin's handover': "Fin escalates conversations based on natural language cues" | Source checked — source checked 2026-10-17 |
| CLM-0033 Intercom suggests branching Fin by audience, for example handing paying customers to a teammate when Fin can't answer while directing non-paying customers to the Help Center. | Documented practice | Use Fin AI Agent in Workflows (Vendor documentation)
Section 'Create different experiences for different audiences': "for paying customers, you might want to hand over to a teammate" | Source checked — source checked 2026-10-17 |
| CLM-0034 Intercom says it tested asking customers for more information before handover extensively at Intercom and observed an increase in answer rate, confirmed resolutions and CSAT. | Self-reported outcome | Use Fin AI Agent in Workflows (Vendor documentation)
Section 'Ask for more information before handover': "We've tested this feature extensively at Intercom and observed an increase" | Source checked — source checked 2026-10-17 |
Cite this record
We Case Study. "Intercom: AI agent answers first, then triages or hands off to the human support team, with configurable handover paths by audience and channel". Decision Lab practice record, version 1. https://wecasestudy.com/decision-lab/practices/intercom-fin-human-handoff
Please cite the original company sources listed above for the facts themselves; this record is a structured, scoped summary of them.
