Amplitude · AI product · Mid · 45 to 60 minutes

Where should AI enter the Amplitude workflow?

Amplitude is a digital analytics platform that helps companies understand user behavior and improve their products. Answer this product management case and get it graded line by line against a weighted rubric.

product sense
experimentation
user segmentation
metrics
tradeoff analysis

The situation

You are a PM on Amplitude's core product experience team. Leadership has asked your group to define how an AI assistant should first show up inside the product. The assistant could help with common analytics tasks and could also play a role in onboarding, but the team is split on whether to add a general chat box, place entry points in specific product surfaces, or focus first on the new user experience. Why now: the company wants to ship an initial AI experience soon, and design, engineering, and go to market teams need a clear plan for what to test first. The hard part is that Amplitude serves both new users who need help getting started and existing users who already have established ways of working. A placement that gets attention might also interrupt analysis workflows or feel bolted on. A placement that is less disruptive might be ignored. You need to recommend where to introduce AI first, how to segment the audience, and how to test whether the chosen entry points actually help users complete meaningful tasks rather than just generate curiosity clicks.

What to produce

Write a 1 to 2 page product strategy memo, about 700 to 900 words.

Your answer must

Identify the primary user segments for this decision, including at minimum new users and existing users, and describe the main job each segment is trying to do when they would encounter AI.
Recommend the 2 to 3 product surfaces or moments you would test first for the AI entry point, and explain why these are higher priority than alternatives.
Lay out an experiment plan with clear hypotheses, test variants, success metrics, and guardrail metrics for each proposed surface.
Explain your onboarding approach, including whether you would test AI as a first run experience, a guided prompt flow, an empty input, or another pattern, and why.
Describe the main risks, especially workflow disruption, low trust, and false signals from novelty, and how your plan would reduce or measure them.

Constraints

Stay within the existing Amplitude product experience, do not assume a separate standalone AI product.
You can propose only a small first wave of experiments, not a broad redesign of every surface.
You must account for both new users and existing users in your plan.
Assume the team has one quarter to run tests and needs a recommendation that can inform an initial launch decision.

Format

Use sections titled: Goal, Users and Jobs, Options Considered, Recommended Tests, Metrics, Risks and Decision Rule.

How Amplitude actually played it

An Amplitude product designer shares how the team tested where to place an AI chat entry point, surface by surface, for new and existing users. Findings include that prompts work only when anchored to user intent and that existing users stick to their habits unless the new path is the only thing in front of them. A practical case on experimenting with AI placement and onboarding.

Test AI entry points by surface and by user segment rather than adding a generic chat box.
Anchor prompts to what the user is trying to do.
Existing users default to habit, so the new path needs to be prominent to be tried.
For onboarding, check whether an empty chat box beats a guided tour.

Built from Your AI chat box isn't a strategy. Where you put it is. by Amy Lin, Amplitude.