Intercom · AI product · Mid · 45 to 60 minutes

How should Fin handle support email?

Intercom is a customer service software company that helps businesses support customers across channels like chat and email. Answer this product management case and get it graded line by line against a weighted rubric.

Product strategy
Prioritization
AI UX
Metrics
Risk management

The situation

You are the PM on Fin AI Agent at Intercom. Fin already helps businesses answer support questions in live chat, and customer conversations suggest a growing ask: can the same AI agent help with support email, which remains a major channel for many support teams. Leadership wants a recommendation on whether to invest now and what an initial release should look like. The opportunity is attractive, but email changes the problem. Threads are asynchronous, messages are longer, one email often contains multiple questions, interactive chat patterns do not transfer well, and deliverability and rendering depend on external mail clients. There is also a trust issue: many end users are less used to talking to an AI agent over email, so the experience must feel natural while making handoff to human support clear. Intercom already has an email product and cross-channel automation foundations, but not a proven AI email experience. Assume you have one squad for one quarter to validate the direction and define what success would look like.

What to produce

Write a 1 to 2 page product strategy memo.

Your answer must

State your recommendation on whether to pursue Fin for email now, later, or not at all, and specify the first customer segment or use cases you would target.
Compare at least two product approaches for bringing Fin to email, including one more conservative option, and explain the tradeoffs.
Define an MVP that covers the end user experience, the AI behavior, and the human handoff flow, and name what you would leave out initially.
Propose a measurement plan with primary success metrics, guardrails, and a learning plan for an alpha or beta, including how you would distinguish answer rate from automatic resolution rate.
Identify the top risks across technical, UX, and operational areas, and give practical mitigations or fallback plans.

Constraints

Build on Intercom's existing email and automation foundations rather than assuming a brand new platform.
Email clients and inbox providers control rendering, throttling, and blocking, so the design cannot rely on perfect delivery or a consistent UI.
The solution must support clear escalation to a human agent for cases the AI should not handle.

Format

Use headings: recommendation, customer problem, options, MVP, metrics, risks.

How Intercom actually played it

Intercom explains how it extended its Fin AI agent from chat to email, where conversations are slower and messier. It reports Fin processed over a million emails in the first month, gave an AI answer in 81 percent of email conversations and resolved over 56 percent automatically on average. A practical case of adapting an AI product to a second channel.

A new channel changes the shape of the problem, not just the UI.
Track both answer rate and automatic resolution rate.
Early customer results help validate a channel expansion.

Built from Fin over email: How we built a multichannel AI agent by Julia Godinho.