Intercom · AI product · Senior · 45 to 60 minutes

How should Intercom reshape support for AI?

Intercom makes software that helps businesses communicate with customers and run customer support. Answer this product management case and get it graded line by line against a weighted rubric.

Strategic thinking
AI product judgment
Prioritization
Cross-functional leadership
Metric design

The situation

Intercom leadership has asked you, a senior PM in the support business, for a recommendation on how the company should respond to rapid progress in generative AI. Customers are showing interest in AI for support, and competitors are beginning to add chatbot style features to their existing products. Intercom already serves support teams that care about fast setup, brand control, and reliable answers to end users, so a weak AI experience could hurt trust. The core question is bigger than whether to ship an AI feature. Should Intercom layer AI onto current workflows, or should AI change the shape of the product itself? A lighter approach would be faster and less disruptive, but might create a disconnected experience and average output. A deeper rethink could create more value, but would affect roadmap priorities, design practices, engineering work, and how teams judge product quality. Leadership wants a clear point of view for the next 12 months, with a plan for sequencing, customer impact, and risk. Assume you have one planning cycle to set direction and cannot fully fund multiple paths at once.

What to produce

A 2 page product strategy memo.

Your answer must

State the strategic choice you recommend, bolt-on AI, a broader product redesign around AI, or a staged hybrid, and explain why.
Define the primary customer problems and target users you would serve first, and explain why they are the right starting point for Intercom.
Lay out a 12 month sequence of major bets across product, design, and engineering, including any changes to how teams work together to build AI features.
Define a minimum quality bar for AI outputs before broad launch, including the metrics, review process, and launch gates you would use.
List the main risks and tradeoffs, including user trust, product complexity, and disruption to existing customers, and explain how you would validate your plan early.

Constraints

You cannot assume exclusive model access, proprietary model breakthroughs, or special data partnerships.
Intercom must continue serving existing support teams, so the transition cannot make current workflows materially worse for customers in the near term.
AI outputs can be wrong or off brand, so your plan must address trust, safety, and consistency from day one.

Format

Use sections such as context, options, recommendation, sequencing, quality bar, risks, and metrics.

How Intercom actually played it

Explains how Intercom treated AI as a forcing function to redesign its whole product around its Fin agent, rather than adding a bolt-on chatbot like competitors. It also notes changes in how designers work with code.

Redesign the product around AI instead of adding a feature
Close the gap between design and engineering workflows
Set a quality baseline for AI outputs

Built from From Stalled to $400M: Eoghan McCabe’s AI-First Design Bet by James Bickerton (Your Product Sucks).