Claude, Figma, Claude, Figma, coffee: how AI really fits into my design work

A tool, and occasionally a window

Most days, AI saves me time on the parts of the work that never needed my full attention: drafting a structure, turning notes into a first version of a document, testing a layout idea in a clickable prototype before I commit to it in detail. Things that used to take an afternoon now take an hour, and the afternoon goes back to design.

Now and then it does something more interesting. A question phrased in a new way, a pattern from a field I would never have looked at, an unexpected angle on a problem I thought I understood. In those moments it works less like a tool and more like a window, opening a view I might not have found alone. I value those moments, but I don’t mistake them for judgement.

What it cannot replace

AI has read a great deal, but it has never sat in a meeting where a client finally admits what is really worrying them. It has never watched someone struggle with a form at a counter, or felt the silence after a proposal lands badly. My view of the world comes from those moments, from years of real projects with real people, and that view is what decides whether a screen is right. AI can help me get to the decision faster. It cannot make the decision for me.

Where clients feel the difference

The biggest change is not inside my process but in how quickly clients can decide. When an idea can be shown as a working prototype the same week, the long debates about what something might look like simply end. People click, react and choose. A decision that once needed three meetings now needs one.

That saves the client money, plainly and measurably. And clients who save money with you remember it. They come back, they recommend you, and the relationship becomes one built on trust rather than invoices. In that sense AI brings us straight back to the oldest story in this profession: long partnerships are earned by being genuinely useful, and anything that makes you more useful strengthens them.

Not everyone can do everything now

There is a popular idea that, with AI, anyone can now do anything. It is simply not true. Anyone can produce something, and quickly, but producing is not the same as solving. A generated interface can look finished and still miss the point of the product, the constraints of the market and the habits of the people who will use it.

Yes, a great deal of average work is being produced right now, and some of it will be good enough for a while. But a market flooded with the average eventually starts looking for the opposite: work that is considered, specific and clearly made by someone who understood the problem. That is where experience earns its keep, and where AI, used well, makes it go further.

Where it doesn’t fit at all

There is a part of my life that AI does not enter, and I don’t expect it ever will. I still draw, paint and make things with my hands, slowly, with all the mistakes that entails. Nothing about that is efficient, and that is exactly the point. It is where I think without a deadline, notice how light falls on a surface and remember what a line feels like before it becomes a vector. No tool can do that for me, because the value lies in the doing itself. I suspect it is also why my screens look the way they do.

So, how does it fit?

It fits between the sketches and the decisions. I use it to move faster, to test more and to show clients something real sooner. Then I close the laptop, look at what we have made and ask the only question that has ever mattered: does this actually work for the people who will use it? No tool answers that one for me, and I wouldn’t want it to.