I Got the Result. I Missed the Way There
by Damiano Giusti
All opinions in this article are my own. They do not represent the views of my employer, and they should not be linked in any way to the company I work for.
Some days ago, someone shared a new AI art tool in a chat at work. Its website had one simple promise: “Start generating artwork immediately.”
A colleague of mine, who is also an artist, was sad about it. Her point was more or less this: that sentence treats art like a productivity task. As if thinking, trying things and discovering were only an annoying step before the result.
When I read her message, I didn’t think about art. I thought about my job. What she described — thinking, experimenting, discovering, and then looking at the result with a big smile — is exactly how writing software feels to me on a good day.
This post is my reply. AI can give me the result immediately. It can’t give me what I learned and what I felt on the way there — and the cheaper results become, the more I understand that the way there was the part I wanted.
When I look closer, the way there gives me two things that the result alone doesn’t. The first is understanding. The second is meaning.
Understanding lives in the process #
You don’t really understand something because it works. You understand it because you fought with it until it worked.
I learned this in 2020, on a tablet app for medical operators, used in ambulances and rescue helicopters. One task was to connect the app to external defibrillators. The vendor gave us an SDK written in C. No Android library, no clear documentation of the protocol. Only C code.
So I read it. I studied the structs, one by one, and I learned how they were serialized into bytes. Then I rebuilt that logic on the Android side, calling into the C code through the JNI. It took me almost four weeks. Most of that time was frustration: wrong bytes, wrong order, a device that didn’t answer. And then, one day, it answered.
I still remember that feeling. Not because the code was perfect, but because I had understood something that was closed to me a month before. That understanding is still with me today.
Now compare it with something I built recently. I designed a feature to keep setlists and a metronome in sync between devices, for a band playing live, over the local network. It’s the same kind of problem: a protocol between devices. Some years ago it would have taken me weeks — discovery, timing, error handling.
With an AI agent, it worked very quickly. And it was while it was working that I realized something. I don’t really know how those devices talk to each other on the LAN. I don’t know how errors are handled. I only see the result in the UI — the UI I imagined.
Same kind of problem. Opposite experience. This time I skipped the four weeks, and I got the result without the understanding. If I had spent the time and the brain on it, I would have learned new things. I didn’t.
That’s not a disaster. The feature works. But it’s a small loss, and I can feel it.
What costs nothing means less #
The second thing the process gives me is harder to explain. It’s what makes a result feel like mine.
I play in bands, and for a long time I wanted a better tool to write and share chord sheets. So I built one, as a side project, by hand. This was before AI coding agents existed. I remember the day I wrote my first real chord sheet with my own editor, exported it and sent it to a friend. Nothing special happened: a friend received a PDF. But I felt an enormous pride. That small file was the result of many evenings of thinking and trying.
The same app is still alive today, but now I build it mostly with AI. It grows much faster, and I’m happy about that. But new features don’t give me the same feeling as that first export. They cost me less, so they mean less to me.
I felt the same thing, much stronger, the first time I tried Suno, the AI that writes complete songs from a short text. My first reaction as a musician was: this is useful. It can quickly make music for a social media post, for example. My second reaction came a few minutes later: we are in trouble as musicians. This thing can replace a lot of what we do.
The numbers are big. According to investor documents reported by Billboard, Suno users create about 7 million songs per day — that is the company’s own number, not an independent one. And in July 2026 the streaming service Deezer said that, at the peak in June, more than half of all new tracks uploaded every day were AI-generated.
But the same Deezer report has a detail I find beautiful. Those AI tracks get only 1–3% of the streams. We produce a lot. We listen to very little of it.
I don’t want to read too much into one number. Still, to me it says the same thing as my chord sheets. What costs nothing to make is easy to make — and also easy to ignore. For the maker, and for the listener.
The other side: the result is often what matters #
Now let me argue against myself, because the other side has strong points.
For an artist, the process is the work. For a developer, very often it isn’t. When a band plays live, nobody cares how I wrote the sync protocol. They want the metronome to click at the same time on every device. The process is a cost. If AI makes the cost smaller, that’s simply good.
And I have a personal example of this too. For years I wanted an app to change the pitch and the tempo of songs, so I could rehearse a full band setlist, from start to end, with all the key changes. No app I tried ever did it the way I needed. With AI, I built it very quickly. Then I liked it so much that I pushed it further and published it: Downstep.
I felt satisfaction. Almost fulfillment. And without AI, honestly, this app would not exist. It would still be an idea in my head, like it was for years.
So the “result immediately” is not always a loss. Sometimes it’s the only way an idea becomes real.
I accept all of this. But it doesn’t change what I love. I love the process. Downstep made me happy as a product. The defibrillator made me happy as an engineer. They are different kinds of happy, and I don’t want to lose the second one.
Laundry and dishes #
There’s a sentence by a writer that went around the internet in 2024, and I think about it often: “I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes.”
I feel the same about my work. There is a lot of boring work in any job: reports, forms, status updates, the same small tasks every week. That’s where I would love AI to take over. Not the part where I sit with a hard problem, try, fail, and finally understand.
I’m not saying people who want results fast are wrong. We are all more impatient than before — me included. I use AI every day, and I skipped the process myself more than once in this post. The question is not if we use these tools. It’s where we point them.
And here I really don’t have an answer. Code, art, music production — they are all crafts. They were built on slow, personal, often frustrating processes. Now the result is a few seconds away for everyone.
So what happens to a craft when nobody needs to go through the process anymore? Does it disappear, does it become a hobby for the few who still enjoy it, or does it turn into something new that we can’t see yet?
I’d really like to know what you think — especially if you make things, with code, with a brush, or with an instrument.
Written by me and my AI assistant :)