Memorisely
Product Management

Frame the problem before you prompt

Slow down before you prompt to make better product decisions.

Zander Whitehurst

10 min read

The fastest way to waste time with AI is to ask it for a solution before you understand the problem. It will happily give you one. It will even look convincing. It just might not be the right thing to build.

I have shipped features I was absolutely convinced were brilliant that turned out to be solutions to problems nobody had. One of them took three months. It had beautiful animations. Nobody used it. That experience taught me more about strategy than any book, and it is why I now get a little twitchy when I see someone open an AI tool before they have written a single sentence about the problem.

The designers and PMs making the best decisions right now use AI to think more clearly, not just to produce more. This article walks through a simple framework for doing that, from defining the outcome to validating your riskiest assumptions with evidence.

Why speed can hurt

AI has made it almost free to generate ideas, screens and even working prototypes. That is a huge advantage, but it comes with a trap. When producing something is easy, it is tempting to skip the thinking that decides whether it should exist at all.

Teams end up with more output and the same number of good decisions. Worse, a polished prototype can create false confidence. It looks finished, so people assume the idea behind it has been tested. Often it has not.

The answer is not to slow everything down. It is to spend a small amount of time framing the problem properly so that all that speed goes in the right direction.

Start with the outcome

Before opening any tool, write one sentence that describes what should be different for the user or the business if this work succeeds. If you cannot write that sentence, you are not ready to design yet.

Good outcome statements are specific and observable. "Improve onboarding" is too vague. "More new users complete their first project within a day of signing up" is something you can design for and measure.

A specific, observable outcome becomes the filter for every design decision that follows.

This single sentence becomes the filter for every later decision. When you are choosing between two directions, ask which one moves the outcome more.

Understand the current reality

Next, gather what you already know. Support tickets, research notes, analytics and sales conversations are full of signals about what is really going on. Most teams have far more evidence than they use.

This is where AI is genuinely helpful. You can paste in messy notes and ask it to group them into themes, pull out direct quotes and highlight contradictions. What used to take a day of affinity mapping can take twenty minutes.

Stay critical, though. AI summaries can smooth over the surprising details that matter most. Always skim the raw material yourself before trusting the synthesis.

Use AI to widen, then narrow

AI is excellent at expanding your thinking. Ask it to:

  • Synthesise research notes into themes
  • Suggest assumptions you might be making
  • List the ways this idea could fail
  • Describe how different types of users might react

Then do the human part. Decide which themes matter, which assumptions are riskiest and which bets are worth testing first.

This pattern of widening then narrowing is the core of good strategy. AI makes the widening step much faster. The narrowing step is where your judgement, context and understanding of the business come in, and that is not something to hand over.

Make your assumptions visible

Write your riskiest assumptions down and rank them by how much damage they would do if wrong. The top one or two are what your first prototype should test, not the whole feature.

A simple way to rank them is to ask two questions for each assumption. How confident are we that this is true? How bad would it be if it were false? The assumptions you are least sure about and that would hurt most are the ones to test first.

Ranking assumptions by uncertainty and damage shows exactly what your first prototype should test.

Making assumptions visible also helps the team align. People often disagree about a solution because they are holding different assumptions without realising it. Writing them down turns a vague debate into a specific conversation.

Frame several options

Before committing, sketch at least three different ways to reach the outcome. They do not need to be detailed. A sentence or a rough wireframe for each is enough.

Comparing options side by side exposes trade offs that a single idea hides. It also protects against falling in love with the first solution that comes to mind. AI can help generate these options quickly, but make sure at least one of them comes from your own understanding of the problem.

Validate with evidence

A quick prototype in Figma or code, shown to five real users, will tell you more than a week of debate. AI makes building that prototype fast. Your job is to make sure it answers the right question.

Design the test around your riskiest assumption. If you believe users want to export their data, do not just show them an export button and ask if they like it. Give them a realistic task and see whether they reach for it on their own.

Write down what result would change your mind before you run the test. It keeps everyone honest when the results come in.

Deciding what would change your mind before the test keeps the results honest.

If there is one habit I would love every designer to steal, it is this: before any prompt, write the outcome down and stick it somewhere you can see it. On a sticky note, at the top of your file, wherever. It sounds almost too simple. It has saved me from building the wrong thing more times than I can count.

Share the reasoning, not just the result

When you present your work, lead with the outcome, the evidence and the assumptions you tested. Then show the design. Stakeholders who understand the reasoning are far more likely to support the decision, and far better at giving useful feedback.

Good strategy is not about having more ideas. It is about choosing well, and AI gives us more time to do exactly that. We practise this kind of framing on real briefs in our AI Product Management Bootcamp.

Keep reading

View all
View all

Optimize your AI for better experiences, not AI slop.

View all

AI Prototyping Skills

Prototype with AI using clean UI patterns and guardrails so flows feel product-grade, not generic template slop.

$99

AI UI Motion Skills

Motion and micro-interaction patterns that make AI-built UI feel designed, polished, and trustworthy in real products.

$99

Fix AI Slop Skills

Spot AI slop early and fix type, spacing, and components so your product looks intentional, not like a template.

$99

AI Responsive Design Skills

Responsive layout patterns for rails, sidebars, and breakpoints so AI-generated UI adapts cleanly from narrow to wide.

$99

AI Dark Mode Skills

Dark theme tokens, contrast, and surfaces that keep AI-generated UI readable and product-grade, not flat gray on black.

$99

AI Color Token Skills

Neutral and primary ramps with clear steps so AI palettes stay systematic, not a random set of accent hex codes.

$99
View all