#4 AI is replacing skills, while the world is losing taste and craft

I have always been obsessed with user experience and product design.

I still remember getting my first iPhone 4s as a kid.

I spent months simply marveling at the level of detail and craft that went into the design.

It was something different from anything I had ever experienced.

Back then I couldn't explain it to myself.

But the border radius of the device chassis somehow looked better than any of the competitors.

The antenna lines were symmetrically aligned.

The skeuomorphic interface was cute and fun to use.

The interface was insanely intuitive.

Buttery smooth (by 2010s standards), no instructions needed.

To me, industrial design and UX felt like art. It is truly something to create such intuitive pieces of technology at the scale of billions of people.

As I got older, besides design, I gained interest in software development. It took me all the way to university to actually start doing anything.

But the idea of simply owning a computer and having the ability to build something unique and allow people to use it, at scale, fascinated me.

I started programming with some basic personal projects, and ended up doing first real projects as client work through consulting. It was fun at first.

But, again, I was solving specific client issues.

My creativity was constrained to problems I didn't care about.

The creativity and quality of my output was limited to specified requirements and whatever sum of money was agreed in the development contract.

Mechanically I was doing what I like - designing, programming & solving problems.

But when my mind was not aligned with the task in hand, I started to feel empty about my work.

I was simply hustling for cash (that I also struggled to make sustainably).

In business, numbers are important for sure. I want to make money too, a lot of it.

But I think we have become so obsessed with these metrics that we have lost the craft and purpose in our work.

It seems like everything revolves around big investment rounds and revenue numbers.

Now with AI we are talking about the incredible productivity boosts.

Shipping in days, hours or minutes.

I get it, numbers and money get you attention.

But apart from some obvious industries (e.g. natural sciences, medicine) there is little talk about how to actually do something better with AI; how to improve the quality of the output, multiply our individual taste, how to operate at the new heights of our abilities beyond productivity.

How to build better, beautiful products.

Even before AI, you could sense this obsession with business metrics instead of obsessing about building truly great products.

Half-assed software with good sales departments that produced wonderful metrics (not despising good sales people). Now with AI, even people with fewer skills are beginning to have abilities to produce that mediocrity.

A clarification: I'm very much pro AI (for many things, there are obvious threats I won't touch on here) and use it for basically everything. And even if AI still occasionally feels immensely stupid, we can't ignore the fact of how philosophically different it is to previous technological revolutions.

As Yuval Harari puts it, AI is alien intelligence that analyzes information, makes decisions and invents new things in a fundamentally different way than humans. And it can do this 24/7, without sleep or rest.

To prevent ourselves from becoming obsolete, I think we must focus on how we still maintain the human taste and care in the output, while utilizing the raw power and intelligence of AI.

A more futuristic way to describe this is that we need to merge with technology (something that is more or less already happening with social media and algorithms).

The unique mind, body, and experiences of each human, that shape the individual's taste for creating unique value for the world are something I don't see AI easily competing with us (even if AI's general creativity may surpass humans).

But unless you foster your individual creativity and develop an awareness around your taste, your output will be limited to what anybody else could do.

Personally, if producing anything unique and high-quality, I saw subpar results and productivity gains from the approach of "just tell the AI what to do and it will do it for you."

I just ended up repeating myself 1000 times to the chatbot and iterating again and again. Validating ideas with AI was exhausting due to sycophancy.

But once I started to re-think the whole process of using AI in my work, the game changed.

I gave Claude Code direct access to my projects. I treated AI as an idea generator, smart researcher and executor, assistant for tedious tasks – instead of a god, or mentor.

I systematized my personal taste and ways of working with AI through context engineering and persistent memory rules that carry a project goal, learnings and taste dynamically as you build from project vision to individual tasks, automatically.

Instead of AI guessing what I want and me guessing what AI just did, we started to work with mental alignment.

The goal: AI does the raw execution; human remains the unique thinker and orchestrator.

Currently my process looks something like this:

research → plan → implement → test → iterate

And underlying the loop, we have:

• Context rules & templates that guide the taste, planning and implementation to produce desired outcomes without repeating oneself.

• Task splitting and isolation allows for parallel agents to execute your vision.

• Persistent memory for AI (and you) to learn from unique failures in your project.

• A pause and interaction point for humans in each phase of the loop to ensure mental alignment with AI and human.

• Again, a human in the loop to understand and govern critical processes and ensure the taste and craft of the end results.

With this approach, I'm starting to see AI produce things that are a reflection of my taste and vision, at speed and complexity that would have taken me much more time to learn and build.

It is still a lot of hard work, as anything worth pursuing should be.

What I essentially want to do is to solve issues that feel personal; build intuitive, beautiful products and sell them for masses. And now with AI, systemize my taste to agents and let them handle manual work while I'm looking after the craft.

I want to develop these products on my own terms (solo, or with small, low-ego and high-agency teams) and earn financial freedom.

Not chasing passive income, but an income where I do what I love at scale.

Working hard, but on my own terms, getting paid based on impact, not hours.

Until some super AGI cooks us, and AIs simply get better at following our directions and hallucinate less, I think we are living beautiful times to make what I previously described a reality.

Mediocre & generic becomes the standard anyone can create, and nobody is willing to pay anymore.

But with human uniqueness & awareness, systems thinking and grounded use of AI, we can build amazing, better products with fewer resources. Operating at the peak of our capabilities and re-defining what is possible.

Outsource the work nobody ever wanted to do to AIs.

Focus on our creative craft, taste and self-expression.

Next – Founding Note #5: We should all be doing we what we love, at scale.