Case studies // FixMyPortfolio

Built an AI tool that grades a designer's portfolio and tells them what to fix.

We combine years of experience with the best technology to
build websites and products for companies who want to move faster.

The FixMyPortfolio landing screen, where a designer pastes the link to their portfolio

About

FixMyPortfolio is a free tool from Flux Academy. A designer pastes a link to their portfolio and gets back a score, a breakdown across six categories, a card worth sharing, and a report on what to fix.

Native built what sits under it: the product, the service that produces a grade, and everything that makes a public tool like this safe to put your name on.

Costs real money to serve. That is the whole problem.

Every visitor

Surprise invoices. The most a day can cost was decided before launch.

Zero

The challenge

Most marketing sites cost the same whether ten people visit or ten thousand. This one spends real money every time somebody uses it, and anybody can use it, free. A designer grading their own portfolio and a script grading ten thousand look identical at the door.

It also has to open whatever link a stranger hands it, from Flux's own servers.

So the risk was never whether it would work. It was whether a good week, or one person with bad intentions, could arrive as an invoice nobody agreed to.

The solution

Anything that is not a real person is turned away before it costs a penny, and turned away invisibly. No puzzle, no checkbox: someone landing on a phone decides in two seconds whether this is worth their time, and a captcha there loses more real designers than it stops fake ones.

Under that sit limits on how much any one person can use it, and a ceiling on the day. Above all of it sits a hard cap on spending that nothing in the code can talk its way past, set by Flux, at whatever number they are comfortable with. Everything else exists so the tool never gets near it.

And it cannot be tricked into opening something it should not. A hostile site can answer a request by pointing somewhere else, and somewhere else can be inside the network it is being asked from. Every step is checked, not just the first.

The analysing screen: a beam scanning down a captured portfolio, with the ten character cards turning behind it

A grade that is wrong is worse than no grade, because a designer will act on it. Getting that right took longer than everything above.

The result screen: a character card with the score, the six category bars, and the form that unlocks the full report

The card is the part that gets posted, so nobody can forge one carrying a score that was never calculated, and one that travels a long way costs no more than one that does not.

The results

FixMyPortfolio is live at fixmyportfolio.ai, and Flux can point anyone at it without watching the meter.

The most a day can cost is a number they chose, not one they find out about. The credentials that spend money never reach a visitor's browser, so there is nothing in anyone's hands worth stealing. And a designer gets a real answer rather than a number: six categories, the evidence behind each score, and the one thing to fix next.

Free
What it costs a designer to use, and Flux knows exactly what it costs them
Turned away
What a script pointed at the tool gets, before it has spent anything
6
Categories behind every score, each carrying the evidence for it
On the record Ran Segall Founder @ Flux Academy

A public tool that costs money every time someone uses it, and that anyone can point at anything, used to be a project with a team and a quarter behind it. This one went out fast, and under budget, and the difference is not that anybody worked harder.

Featured Case Studies

We work with teams at every stage, from a first marketing site to a full product, and cover strategy, design, and development as one unit.

omnilab consulting b2b saas 2026

Rebuilt the digital home of a demand generation agency that sells performance for a living.

read case study

trase.ai enterprise 2026

Built the web presence for a governed AI platform sold where failure is not an option.

read case study