Built with AI. Finished properly.
You built with Lovable, Claude Code, Replit or v0. It runs. I tell you where it falls short and hand you the fixes as prompts.
Looks finished. Isn't.
AI quickly builds something that looks plausible. Whether it's also a good product decision, nobody checks. Unclear audience, missing error states, navigation only the founding team understands: users notice right away, you usually notice too late.
Four layers, one outside view.
Product
Is it clear who this is for and what the core task is? Do you get it in five seconds?
UX
Flow, navigation, empty and error states, onboarding, clarity.
UI
Hierarchy, typography, consistency, responsiveness. Does it feel all of a piece?
AI Build
Where did the AI build something pretty instead of something thought through?
Problem. Why. Change. Prompt.
A typical example from an AI-built dashboard. You get every one of your findings in exactly this format.
The first login shows an empty dashboard. Three cards showing “0”, an empty table, no next step.
An empty state without a next step looks broken, not new. The AI built the “with data” state because you described it. Nobody described the state before that.
An empty state with one clear action and optional sample data to try things out.
If a user doesn't have any projects yet, show an empty state on the dashboard instead of the empty cards: heading “Create your first project”, one sentence on what the dashboard will show afterwards, a button “Create project” and a link “Try it with sample data”. Use the existing components, change nothing else.
How the check works.
You tell me
In the form: what you built, for whom and where it falls short.
I plan
Scope and date within two working days. Only then do you pay.
We talk
In the live call I show you everything and we set the order.
You have it in writing
After one week: findings, prioritised, each with a prompt.
Better one flow done right than everything a little.
Everything
Up to about 10 screens, I look at the whole product.
The most important flow
For example sign-up to the first success moment, booking or checkout.
Audit to strategy
More flows, the whole MVP, a redesign or the product strategy. We agree on the scope before you pay.
Hey, I'm Cornelia.
UX and product design for 15+ years. I build with AI every day. I know the points where AI-built products tip over, because that's how I build myself.
One check. Prioritised findings. Ready-made prompts.
You only pay once we've agreed on the scope together. No subscription, no surprises.
- A small app in full or the most important flow
- Pre-review and a 40-minute live call
- Findings, prioritised, with ready-made prompts
- Written report within one week
- Recording, if you like
- Money back if I find nothing
- Everything in the AI Product Check
- Several flows or the whole app
- Follow-up call after two weeks: we review the implementation together
Prices plus statutory VAT where applicable
Tell me about your product.
A few details are enough. I'll get back to you within two working days with scope, date and invoice.
Before you ask.
What do I need to prepare?
Just the request form. If a login is needed, we'll sort out test access by email afterwards.
My product is bigger. Is the check still worth it?
Yes. Then I look at the most important flow, the path your product stands or falls on. For more, there's the audit. I'll tell you what fits after your request.
What tools does it have to be built with?
Doesn't matter. Lovable, Claude Code, Replit, v0, Bolt, Cursor or built the classic way. I write the prompts so they work in your tool.
Do you also check code, security or performance?
No. The check looks at product, UX and UI, so at what users experience. For security and code quality you need a technical review.
My product isn't public yet. Is that okay?
Yes. Everything stays confidential, and I'll sign an NDA beforehand if you like.
What if you find nothing?
Then you get your money back. If I find nothing worth changing, I'll tell you in the call and refund the amount.
Is the call recorded?
Only if you agree. Only you get the recording.
And after that?
Many implement the prompts themselves. If you want more: an audit for the whole MVP, a redesign or the product strategy. Or we solve it together in a sprint.