AI prototyping practice exercise for product managers

What is the challenge scenario?

It is 8:47am on Thursday. You have a design review at 10:00 with your PM lead, two engineers, and a stakeholder who specifically asked to see the onboarding flow for Trackr, a project management tool for small agencies, before the sprint kicks off. Your junior designer built a first version overnight using Lovable and sent it over at 8:30 with a note that a few things feel off.

They are more than off. The CTA on screen two says Continue but actually submits a form with no way back. Screen three shows a populated task list even though most new team members have zero tasks assigned on day one. The dashboard's empty state on screen four is a blank white rectangle. Every button across the flow says either Get started or Continue, with no copy specific to what is actually happening. And the mobile view of screen four is broken, with the sidebar overlapping the main content, which matters because the stakeholder is joining from their phone. Each fix takes about 15 minutes with Lovable. You have 73 minutes before you need to stop and prep. You can fix four. One stays broken.

What are you asked to do?

Choose which four of the five problems you would fix and the order you would tackle them in, writing one sentence for each on why it makes the cut, and one sentence on why the problem you are leaving broken is the lowest priority for this specific meeting and audience. Write the exact revision prompt you would give an AI prototyping tool for each of your four fixes, stating the current behavior, why it is wrong, and what correct looks like. Rewrite the copy for all four screens so every headline, button and empty state speaks to a new, slightly anxious team member with three minutes to spare. Close by writing the two or three sentences you would say at the start of the review, before anyone clicks through to the broken part, naming the issue honestly and saying when the fix will land.

What does a strong answer cover?

This bootcamp does not publish a rubric for its practical challenge, so check your own draft against the curriculum's own questions instead of a worked solution:

• Does the triage choice reflect the specific meeting and audience, rather than simply picking whatever is easiest to fix?
• Does each revision prompt state the current behavior, the correct behavior and one specific constraint, rather than a vague request?
• Does the rewritten copy sound like it was written for a new, non technical team member, not generic onboarding filler?
• Does the plan for the broken screen name what the fix looks like and when it will ship, rather than avoiding the topic?

Skills this exercise practises

Triaging prototype issues under a real time constraint and a specific audience, writing precise AI revision prompts that fix one thing without breaking adjacent screens, writing functional onboarding copy, and communicating honestly about an incomplete artifact.

Which bootcamp does this challenge come from?

This is the practical challenge from Prototyping with AI, Builders Camp's one week bootcamp on using AI to accelerate ideation, UX flows and prototype creation, led by Andre Albuquerque across two 90 minute live sessions. Completing it counts toward that bootcamp's certificate. The how to build a prototype with Lovable guide covers the tool this exercise is built around in more depth, and best AI tools for product managers and ChatGPT for product managers both help with the prompt writing side. The prototype review practice exercise elsewhere in this series practises the same critique skill without the AI fix step.

Builders Camp runs live and self paced bootcamps in product management and AI product building. See the Prototyping with AI bootcamp for current cohort dates and the full self paced curriculum.