I didn’t expect this to be the first thing I’d post on Substack. But here we are.

I had a gap between two contracts. Not a lot of time to fill it with something meaningful, but enough to feel restless. And I thought… I HAVE skills. The kind I use to help brands find their positioning, improve their experience, sharpen their digital presence, create better processes.
So what if I turned them on myself for once?

The pain that I knew was coming

I started thinking about the problem I’d be facing in the next few weeks. And I found one.

Heuristic reviews for UX audits.

I love the analysis, I genuinely do. Walking through a website, identifying friction, spotting what’s broken or inconsistent : that’s the good part. What I’m less fond of is everything that comes after. The screenshots, storing them, dragging them into Figma, resizing, aligning, annotating. Making sure every sticky note has the same spacing, the same base, the same visual logic. Formatting a report that looks clean enough to actually be read by someone else.

I have always preferred thinking over formatting. And every time I’d go through this process, I’d feel this lightly annoying productivity dislike. Not because the work wasn’t valuable, but because I was spending my thinking time on formatting instead of insights.

So I decided to automate the part I hated.

Thinking it. Building it. Testing it. Iterating it. Repeat

I didn’t write the code from scratch. I vibe coded with Claude, which, if you haven’t heard the term (I’m sure you have by now), means you have an idea, you talk to an LLM and together you build something that matches your vision. You’re directing. You’re describing. You’re testing. You’re not pretending to be an engineer.

My first idea was simple: paste a URL, get the website to load, take a screenshot, annotate it, add findings and recommendations, export a report. I uploaded my use case along with some insight about… me. That was it.

But the first issue was that some websites don’t allow themselves to load in an iframe. That approach fell apart quite fast.

So I decided to add a Chrome extension. It would capture the website and send the image straight into the app, ready to be added to the report. That worked really well.

The pivot

Then I had a conversation with a creative director I work with and respect. He told me the PDF format wasn’t really suited to what we’d be doing the following week. He’d rather have something usable in Figma. I agreed. So I went back in.

I was a bit sceptical about creating a Figma Plugin, but I had a bit of time to give it a try.

I figured out how to give Claude detailed instructions about the layout : the structure of the frames, the quality of the images, the format of each annotation, so it would integrate cleanly into how we work at the agency.

And then I hit another wall: there was no clean way to push content directly into Figma automatically. The best path was to export a JSON file and re-import it. Honestly? Good enough.

The output lands in Figma as frames: images, editable text, annotations, all organised. Ready to present. No dragging. No resizing. No copying and pasting into a template.

The numbers now…

After trying it, I was quite happy to see how fast my process was but I wanted a tangible estinate.
When I asked for an analysis of what the tool did well and where it fell short, the numbers came back like this: ≈77% faster.

I spent a day building this during a gap where I wasn’t booked anyway. And now I have something that means I can spend more time on the actual thinking : the strategy, the patterns, the recommendations that matter, instead of making sure my sticky notes are consistently colour-coded.

AI doesn’t do the work for me, but it frees up time so I can focus on the most impactful tasks.

Why THIS matters more than the tool

There’s something I want to say about all of this that feels more important than the tool itself.

I love talking about Experiences with a capital E. The big brand thinking, the cultural moments, the systems that shape how people feel about a product or a physical experience over time. That’s what gets me excited. But this. This tiny, specific, lightly annoying productivity dislike is also part of the work. There are people behind every brand experience. Processes behind every deliverable. Micro-tasks behind every insight.

And when those micro-tasks quietly take up the space where your best thinking and best delivery should live, the big E suffers too.

Vibe coding didn’t make me a developer. I was one, in another lifetime. But it gave me one afternoon back, and that afternoon now belongs to something better.

A tiny tool that solve my extremely niche painpoint and has a huge impact in my day.

Try it yourself

Want to try the tool? if are a UX designer/researcher, I’ve put the link below. There’s a README if you need it. It’s my first time vibe coding, so it’s still a bit rough. I didn’t spend a minute on the UI, and it might break with Chrome or Figma updates, but it’s there, and it’s useful. Give it a try, and let me know what you think.

https://heuristic-iq.netlify.app/

I’d love to see what kind of micro-tasks you’ve automated, or what you’d build if you had a free afternoon with Claude