Figma · AI product · Senior · 45 to 60 minutes

Should Figma bet on prompt to app creation?

Figma is a collaborative design and prototyping platform used by teams to create digital products. Answer this product management case and get it graded line by line against a weighted rubric.

Product strategy
AI eval design
Prioritization
User insight
Stakeholder management

The situation

David Kossnick, Head of Product, AI, asks you for a recommendation before Figma commits meaningful headcount to a new AI surface. The company already has new website publishing infrastructure from Figma Sites, plus internal experiments that can turn designs into code and, in some cases, generate a functional site or small app from a prompt. When the prototype works, reactions are strong. When it fails, it fails visibly, and quality is hard to judge because both look and behavior matter. The team can imagine many AI features across Figma, so the real question is not only whether this concept is exciting, but whether it is viable enough to become a product bet now, later, or in a narrower form. You need to propose how Figma should decide, what user value to optimize for first, how to evaluate output quality before broad launch, and how to widen feedback without confusing novelty for durable demand. Your recommendation will shape staffing, scope, and whether this becomes a central AI initiative or remains a prototype.

What to produce

A 2 page product strategy memo, about 900 to 1,100 words.

Your answer must

State your recommendation on whether Figma should invest now, wait for the technology to improve, or narrow the product, and give the decision criteria behind that call.
Define the initial target persona, the top 1 to 2 user jobs to be done, and what should be explicitly out of scope for the first version.
Design an evaluation plan with 2 to 4 success metrics that cover output quality and real user value, and specify who should judge each metric and why.
Outline a staged feedback rollout from internal builders to target personas to broader internal or external users, including what you would learn at each stage and the gates to move forward.
Identify the main risks and second order effects, including brand quality bar, overlap with existing Figma workflows, and implications for team structure and shared AI infrastructure.

Constraints

Build on Figma's existing Sites and design to code work, not a clean sheet platform.
Early evidence is mostly prototypes and qualitative usage, not mature product analytics.
Output quality is probabilistic and partly subjective, so a single deterministic acceptance test will not be enough.
Your plan must support fast iteration by a small cross functional team.

Format

Use five sections: recommendation, target user and scope, decision logic, evaluation plan, rollout and risks.

How Figma actually played it

Shows how Figma decided which AI ideas to pursue and judged Figma Make with scored human review, model judges and live A/B tests.

Use a clear decision tree for AI bets
Grade outputs with several methods
Widen feedback circles step by step

Built from The Art of Evals: How Figma Put People at the Center of Its AI Product by First Round Staff (interview with David Kossnick).