Notion · AI product · Entry · 45 to 60 minutes

How should Notion AI guide first use?

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

prioritization
product sense
AI UX
metrics
tradeoff analysis

The situation

You are the PM for Notion AI after an early alpha release to a limited group of users. The feature is inside Notion pages and can help generate or transform text, but the team still does not know what the product should be best at first. Some early users open a blank page and want help getting started. Others already have notes, proposals, or meeting writeups and want the AI to improve, summarize, or organize what is there. Leadership wants a recommendation for the next release: when someone triggers AI in Notion, what should the product steer them toward, and what workflow should be most polished? The choice is hard. You could optimize for a few common writing templates, invest in follow up editing and rewriting around generated text, or focus on turning messy workspace content into clearer outputs like summaries or action items. Interest in AI is rising fast, and Notion needs to make AI feel useful in daily work without turning the product into a separate chatbot or a confusing pile of prompts.

What to produce

A 1 to 2 page product recommendation memo.

Your answer must

State the main user problem you would prioritize and the user context you are solving for, blank page creation, working with existing content, or messy team notes.
Recommend one primary product direction for the next release and describe the core in page user flow.
Compare your recommendation against at least two plausible alternatives and explain why they are lower priority now.
Define 3 success metrics, including one adoption metric, one quality or usefulness metric, and one behavior metric tied to what users do after the first AI response.
List the biggest risks and how you would reduce them through UX, prompting, or launch sequencing.

Constraints

The solution must live inside the existing Notion page experience, not a separate chat app.
The next release can meaningfully polish only one primary workflow.
Assume model output is often helpful but not consistently good enough to publish without human review.

Format

Use short headings: user problem, recommendation, alternatives, metrics, risks.

How Notion actually played it

Notion shares what it learned from the first months of Notion AI use. People did expected tasks like drafting but also much more, and more than half of users edit the AI's first draft, with few accepting it as is. A short, data-grounded note on real AI usage.

Users treat AI output as a draft to edit.
Real usage goes beyond the planned use cases.
Design for iteration on output, not one shot.

Built from Millions have used Notion AI. Here's what we've learned, and what you should try by Jonas Lavoie and David Tibbitts.