Notion · AI product · Mid · 45 to 60 minutes

How should Notion add AI to its workspace?

Notion is a workspace app where people write documents, organize knowledge, and manage projects in one place. Answer this product management case and get it graded line by line against a weighted rubric.

Product strategy
AI judgment
Prioritization
Risk management
Metric design

The situation

You are the PM for a core product area at Notion. Senior leadership asks you for a recommendation on whether and how generative AI should be introduced into the product. Interest in AI has surged, users are starting to ask about it, and there is visible market pressure to respond quickly. At the same time, Notion is a mature product used for real work, so low quality outputs or confusing behavior could damage user trust. Your job is not to prove that AI is fashionable. You need to decide where AI could add real value inside existing workflows, what risks matter most, and how much of the first release should be built around safeguards versus speed. The hardest part is that the product already does many jobs well without AI, so a weak or bolted on feature could create noise instead of value. Leadership wants a clear recommendation that balances urgency with product quality and trust, and they want to understand what a sensible first launch should and should not include.

What to produce

Write a 1 to 2 page product recommendation memo.

Your answer must

State whether Notion should launch an AI feature now, later, or in a limited way, and explain why.
Identify 1 to 2 user jobs or workflows inside Notion where AI is most likely to create real value, and explain why these are better than other options.
Propose a scoped first release, including what is intentionally out of scope.
Describe key risks, especially around incorrect or misleading outputs, and give concrete guardrails for the initial launch.
Define success metrics for the first 90 days, including at least one adoption metric and one trust or quality metric.

Constraints

You cannot rely on a broad standalone AI product outside Notion's existing workspace experience.
You have limited engineering and design capacity, so the first release must be narrow.
You cannot assume AI outputs are always correct, and the experience must set the right user expectations.
Do not use competitive parity alone as the main reason to build.

Format

Use sections with headings such as recommendation, target workflows, first release, guardrails, and success metrics.

How Notion actually played it

Draws five lessons from Notion AI's launch for PMs under pressure to add AI, including adding guardrails such as visible disclaimers when outputs might be wrong.

Do not add AI only from FOMO
Bake guardrails into AI features
Study how a mature product integrates AI into its core

Built from Notion AI: 5 Lessons for Product Managers by Behind Product Lines.