No 13: New ways of working

This post has been brewing for five months now. It’s a hard one and I tend to avoid talking about AI. It receives so much attention as it is, but it's useful to see how actual people are using it in their jobs.

It's mostly a collection of notes organized by theme.

I've been painting more on and off. I finally stopped trying to make good paintings and instead paint for fun with my kid.

Sometimes I feel sad that I'm a failed artist. Most of the time though, I'm fully aware that we are all constantly failing at pretty much everything we undertake.

Even AI fails constantly.

The past six months, I've been designing in code.

I've met designers who never touched code before and in 3 months were already ditching Figma. The trick is to stay curious, keep playing with it and learn from other designers who might be ahead of you.

The path to AI fluency is one you walk, picking up skills along the way. It's an act of learning while doing.

When I first got into product design, my mentors were software engineers. Aligning with them was beneficial in getting to shipped product faster.

The collapsing distance between design and shipped product means what exactly?

Engineers giving designers access to pull requests lets us propose direct front-end changes or code updates.

It means building high-polish, working prototypes that ingest real design system tokens and components. I can then deploy to Vercel or GitHub.

It’s all in code and it's faster now to see the full product experience.

We're moving away from single interfaces and considering an increasing number of variations of a product.

It can be hard to focus when there are so many options to consider. One way of staying on track is to outline what AI can't do and prioritize user-centered work.

The real risk is that AI-generated outputs might be coherent, but may not solve any real need.

Designing with AI alone quickly leads to products devoid of resonance.

Design judgement introduces friction between instruction and output.

If an organization doesn't welcome judgement, it gets accelerated product development cycles that eventually stall because nothing new is entering the system.

AI mimics design skills, the same way it mimics empathy. It's not really a good designer.

AI relies on a statistical predictive process, not a creative one. AI is also not grounded in truth and at times not even seeking it.

Designers introduce friction. We push back.

If organizations are replacing design roles with AI, those orgs never understood or valued design to begin with.

We're nearing the end of summer.

I've been wearing a perfume called Creosote, which launched a day after my birthday.

It took over a year for Bruno Fazzolari to develop it. Inspired by the smell and feeling of Southwestern rain, when cool air rolls in with July thunderheads. The scent is not a photorealistic recreation, but the essence of what desert rain feels like.

Five years ago, Bruno moved back to Tucson where he grew up because his elderly parents needed help.

He wrote about the first monsoon that rolled in after he moved back: "The wind would change first. Then suddenly the whole neighborhood smelled green and wet and electric at the same time. Creosote, hot pavement, damp earth."

As I'm writing this, I'm wearing another perfume he developed last year while his father's health was declining. Fazzolari described the process as "my sanctuary as my father declined—a place to process grief while still building something beautiful."

In the bottle, the perfume looks almost black and mildly olive green on my skin. This limited release is called Deixis in reference to being present in a moment.

There are so many people in this world who will never benefit from AI.

It's good to remember that.

Spend some time in ancient forests and old buildings.

Until next time!

Eva