Terms

Before listing features, explain who has the problem, what they need, and why it matters.

Write the refund request end to end: how the customer starts, how the system responds, and what failure looks like.

Map out every step a user takes from arriving on the site to finishing checkout.

Map the whole experience from first hearing about the product to actually using it, including where people get stuck.

This page feels too generic. Which customer should I write it for?

Help me spell out what I'm building for the AI: the problem it solves and what counts as done.

Don't have AI start coding features yet. Let's do product discovery first to see whether users actually care about this problem.

Let's start with the smallest version that proves the core feature works instead of building everything at once.

Put these requests and problems into an ordered backlog, but do not treat every item as a delivery promise.

Turn the next few stages into a roadmap that explains why and what to improve, without scheduling daily development tasks.

Put design, development, testing, and launch on a timeline with dependencies, and show what a delay changes.

Hold off on colors—sketch a gray page skeleton first: header, hero, three cards, footer, just the positions.

Here are my references: the three I like set the feel, the crossed-out one is what to avoid. Style the site this way.

I asked AI to make only the login button blue, but the header and card shadow also changed. How should I review that?

Five pages each have a copy of the button style, and one was missed during a color change. What does that show?

Make a clickable prototype to confirm the flow before writing code.

Don't just say this interface 'feels clunky'. Run through Nielsen's usability heuristics to pinpoint whether status is hidden, error prevention is missing, or terminology is confusing.

Do not choose between these sign-up pages by taste. Randomly split users and compare the same metric.

Turn visits, sign-up starts, profile submissions, and completions into a funnel to find the largest drop-off.

Build me a client quote list: add new ones, edit amount and status, delete rows, and show all records on open.

Each quote should record client name, amount, date, and status—just these four, nothing more for now.

The amount column is stored as text and cannot be summed. Change it to a number type, and switch date to a real date picker.

Add an export button to the quote list that downloads all records as a CSV that opens cleanly in Excel.

I want to know whether anyone actually clicks this button.

AI gave me a regex I cannot read. Should I use it?