AI tools built by designers (not for designers): a curated list

Most AI design tools are built to make designers faster. Generate a component, write some UX copy, export a spec. The underlying idea is the same: reduce friction in the production pipeline.

That’s valuable. But it points designers toward speed, which points them toward the part of the work that’s easiest to compress with AI: and that’s also the part least likely to be their long-term competitive advantage.

The tools worth paying attention to are the ones built for a different purpose: understanding better. Your designs, your users, and the decisions underneath the work.

Here are five built by designers, each packaging genuine design expertise into something reusable.

Layers Skills, by Jamie Mill

Layers Skills is built around a seven-layers framework that covers the full design process from research through to polish. Rather than prompting from scratch each time, you load the skill for the phase you’re in.

The idea is that deliberate process produces better results than ad-hoc prompting. Each skill in the library represents a practised approach to a specific type of design work. You get consistency across your process: not just faster generation, but more rigorous evaluation.

This is what good AI-assisted design practice looks like: a structured system built by someone who’s thought carefully about what each phase of design work requires, made accessible to anyone who needs it.

UXOverflow, by Devendra Tayade

UXOverflow lets you attach Figma designs and specify your audience to get a scored analysis across clarity, friction, and goal alignment.

The key distinction is specificity. Generic AI design feedback answers “what could be improved here?” UXOverflow answers “does this design achieve its goal for this audience?” That’s a much more useful question.

Goal alignment is particularly interesting. An onboarding screen that’s clear and friction-free but doesn’t guide users toward the behaviour the business needs is still a failure. Evaluating against goal alignment rather than just usability captures that.

Gestalt Or Not, by Christina Wodtke

Gestalt Or Not analyses designs against Gestalt psychology and visual design heuristics, with an important extra: it explains the principles behind the feedback.

Most AI design tools flag violations. This one teaches. When it identifies a grouping problem, it explains why visual proximity matters and how the brain interprets it. When it flags inconsistent visual weight, it connects that to how attention flows through a design.

This dual-purpose output, validation and education, makes it genuinely useful for developing design instinct, not just catching problems.

Value Activation Model, by Parker Simon

Value Activation Model translates user research into design decisions using established behavioural frameworks. You give it what an audience values, and it traces that down through emotional and behavioural principles to a concrete design pattern.

This is solving one of the hardest problems in design process: the gap between research findings and design decisions. “Users want to feel confident” is a common research output. What that means for a specific interface element is usually left as an exercise for the designer. This tool makes that connection explicit.

Qualia, by Andrea De Iturbe

Qualia audits Figma prototypes end-to-end: not just static screens, but the interaction logic and user flows underneath them.

Static screen audits miss a significant category of UX problems: the ones that only appear in motion. A design can look perfect in Figma and still have broken flows, confusing state transitions, or friction that only shows up when you actually click through it. Qualia is built for that.

For teams who want to catch broken flows before they ship, this is the kind of tool that sits between design review and handoff.

What these tools have in common

Each of these was built by a designer who had a specific problem they needed to solve: and decided that AI made it possible to package the solution into something others could use.

They’re not replacing human judgment. They’re amplifying it. Each tool embeds a framework developed by a practising designer and makes it accessible without requiring the same depth of background knowledge.

That’s a different model from “use AI to produce screens faster.” It’s using AI to distribute expertise, to make rigorous evaluation accessible at the moment you need it, and to catch the kinds of issues that are hard to see when you’re too close to your own work.

Building your own

The tools above are products. But the underlying approach, building a structured skill around a specific framework, is something any designer can do.

If you’ve developed a rigorous approach to a specific type of design problem, you can package it as a reusable AI skill. The investment pays off when you run the same kind of evaluation repeatedly and want consistency, or when you want to share an approach with your team without requiring everyone to develop the same depth of expertise.

I built a UX audit skill using a stack of established heuristics frameworks alongside custom checks for first-time users, trust at high-stakes moments, and cognitive load. The result is an audit I can run on any screen in a few minutes, with findings that cite specific principles and carry their own justification.

The point isn’t to use AI to avoid thinking. It’s to use AI to think more consistently, more rigorously, and at moments in the process where you’d otherwise move too fast.

What to look for when evaluating design AI tools

Not all design AI tools are built on the same level of underlying thought.

The good ones are anchored in specific, established frameworks. When they give you feedback, there’s a reason that can be traced back to something with evidence behind it. When they identify a problem, they can explain why it’s a problem, not just flag it.

The less good ones apply aesthetic preferences that happen to be common without being able to explain why. “This feels cluttered” is a different kind of feedback from “this violates Miller’s Law: users are being asked to process more than working memory can hold.”

The difference between the two is the depth of design thinking embedded in the tool. That depth is what designers bring to AI, and it’s what makes AI genuinely useful for design practice rather than just fast. The same principle applies when you build your own: I’ve written about communicating design value in benefits, not features, and the tools that stick follow the same rule.