Notion · Execution · Mid · 45 to 60 minutes

How should Notion launch its first AI workflow?

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

Prioritization
Product sense
UX judgment
Execution planning
Metrics

The situation

It is late 2022, and Notion has early access to a much stronger language model than the team has seen before. The CEO, a cofounder, and a small product and engineering group have built a rough prototype during an offsite, and leadership now wants a recommendation on what the first AI experience inside Notion should be. The team believes AI could become important to how people write, think, and find information in Notion, but there is no agreement yet on the best starting point. Options range from writing help inside the editor, to knowledge retrieval across a workspace, to a more general assistant layer. You also need to decide how people would invoke AI, whether the experience should lean on preset actions or open ended prompts, and how the first rollout should work for different customer types. Assume you have four weeks to get to a private alpha. Your challenge is to define a narrow, shippable first version that feels native to Notion without adding confusion or splitting the product into disconnected experiences.

What to produce

A 1 to 2 page product recommendation memo.

Your answer must

Recommend one primary AI workflow to ship in the private alpha, and explain why it is the best starting point versus at least one credible alternative.
Propose the core user experience, including how users trigger AI and how the design balances preset actions with custom requests.
Define a rollout approach that covers whether the feature is enabled by default, who can control access, and how to avoid confusing users.
List the top success metrics for the alpha, plus 3 to 5 specific learning questions the team should answer before broader release.
Identify the biggest execution risks, including trust or quality issues, and give a mitigation for each.

Constraints

The team has about four weeks before private alpha, so scope must be tight and realistic.
The first version must live inside Notion, not as a separate standalone product.
The core editor and existing workflows are already widely used, so the AI experience cannot make the product feel much more complex.
Model quality is promising but imperfect, so the launch plan must protect user trust.

Format

Use five short sections: recommendation, target users and alternatives, UX, rollout and metrics, risks and mitigations.

How Notion actually played it

Notion's CEO traces how an offsite prototype became a launch, including the writing focused interface, spacebar entry point and default on setting.

Prototype quickly with the whole team
Choose entry points people will find
Avoid forking the product into two versions

Built from The crucial moments and decisions leading up to the launch of Notion AI by Ivan Zhao.