This morning, while I was making breakfast, I built a production-ready digital product.
Not a prototype thrown together just because, not a mockup with the implicit “we’ll fix it later.” I mean something actually alive, with a domain, a working backend, a usable interface. Something a real user can open right now and use.
It took me 17 minutes. I didn’t write a single line of code. I talked to an AI tool.
The result is music-map.uk, an app that answers a question you may have never asked yourself: what does this place in the world sound like?
And no, this post isn’t about music-map. Or rather, it isn’t about the product itself. It’s about what it means that something like this can exist like that, in that way, while the water for coffee is coming up.
The distinction that changes everything #
Here I need to clarify something, otherwise it sounds like I’m telling a story that’s already existed for years, just with a bit more excitement.
There are consumer tools that promise exactly this: open a browser, describe the idea, and in a few clicks you have an app. Lovable, Bubble, Glide, Softr. The ecosystem is huge and keeps growing.
The point is that, today, there’s still a pretty clear gap between “I built something that looks like an app” and “I built something that holds up in production.” I’m not saying this out of snobbery, and not as a judgment on those products either. In certain use cases they’re genuinely great.
It’s really a matter of substance: robustness, scalability, control over the generated code, the ability to intervene when something breaks, real ownership of the infrastructure. That very concrete feeling of knowing where the pieces are and what happens if one of those pieces stops working.
I used tools meant for people who, in some way, already know what they’re building. Tools that assume you have technical context, that you can read an error, that you can make architectural decisions when they put them in front of you.
That difference is still enormous today. And that is exactly the point.
b2b doesn’t scare me, at least for now #
While I was finishing my coffee, the question that always hangs in the industry hit me, even when we pretend it doesn’t: is this taking my job?
In b2b the answer, as I see it right now, is no. Or at least, not in the catastrophic sense that mainstream narratives love.
Because good b2b work was never “writing code.” It’s understanding a client’s context, translating often messy needs into precise technical requirements, making sure a system holds up under pressure, navigating compliance and regulations, maintaining relationships of trust over time.
In day-to-day work at Oltrematica, I often have on the table at the same time a migration from Python to Laravel 12, compliance frameworks for the Cyber Resilience Act, sbom platforms for clients who have to report their software to public bodies, and solutions tied to the Product Liability Directive that will come into force in 2026. That’s not stuff you solve with a 17-minute conversation.
AI has changed how I do that work. The speed on certain tasks has increased in an almost embarrassing way. But the value I brought wasn’t in typing the code. It was in knowing what to write, why, and above all what to avoid.
That part, for now, hasn’t disappeared. If anything, it may have gotten stronger, because it leaves me more room to think.
Consumer is a different story #
On consumer, instead, I feel less calm. Not because apps disappear tomorrow, but because it seems to me the shape of the market is changing.
If you take the classic adoption curve—innovators, early adopters, early majority, late majority, laggards—and ask where we are with ai tools for developing software, something weird comes out.
For developers, we’re already in the early majority. Many use them every day, integration into workflows is real, productivity is measurable.
For a non-technical user, instead, we’re still between innovators and the very first early adopters. The technical threshold is still high. Not as high as ten years ago, but high enough to exclude most people.
And that threshold drops every month. Not always linearly. Sometimes there’s a jump, maybe because someone finds an interface that really works, or because a model becomes capable enough to handle the ambiguity of natural language without forcing you to correct it every two minutes.
When that threshold drops enough to be crossed by a normal person—someone who doesn’t know what a server is and doesn’t even want to know—the consumer software market will change irreversibly.
The real “product” of consumer apps #
Let me try to say it simply.
Over the last fifteen years, a consumer app worked like this: someone had an idea, and only those with technical skills, or money to buy them, managed to turn it into a product. Then that product was distributed and used by others.
The value was in execution, sure. But underneath there was something else: distance.
The distance between having an intention and being able to use it.
That gap, between “I’d like to” and “I can,” has been the market. Consumer apps existed because there was a technological barrier between those who knew how to build and those who wanted to use.
If that distance collapses—if anyone can describe what they want and receive in return working software calibrated to their needs—what remains of the traditional model?
I’m not saying apps disappear tomorrow. I’m saying the mechanism that has supported a huge part of consumer could stop working sooner than we expect.
Will custom software become normal? #
For a few months now, an idea has been buzzing in my head that feels radical, but maybe it’s only radical because we’re used to the opposite.
In the physical world, mass production makes sense because personalization costs. A custom chair costs much more than an Ikea chair. So you accept a compromise and buy something standard, “good enough.”
Software has worked the same way. A note-taking app is designed for millions of people, so it will make choices that work for many and are perfect for very few. You buy it because building it custom, until yesterday, cost years or thousands of euros.
But what if the cost of personalization collapses to almost zero?
If I can describe how I truly want my notes app—with the categories I use, the flow I prefer, integrated with the tools I already have—does it still make sense to buy the one designed for millions of users?
Maybe yes, for convenience. Maybe no, if the difference between “adapting” and “having it the way I want” becomes too obvious.
I wonder if mass-market software will end up like commercial music in the streaming era. It remains, sure, but it stops being the center of everything, because alongside it grow more personal, more tailored experiences.
Who might resist #
If this scenario comes true—and I’m putting an “if” not because I doubt the direction, but because the speed and shape of the change are truly uncertain—there are consumer models that seem more solid than others.
Network platforms, for example. Twitter, Linkedin, WhatsApp. Their value isn’t in the app itself, but in the fact that everyone else is in there. You can’t have your own customized version of a network, because a network without the network is just an empty interface.
Then there are services with proprietary data. Spotify doesn’t sell a music player. It sells access to catalogs, metadata, licenses, algorithms fueled by billions of listens. You don’t generate that stuff with a prompt.
And there are products where trust and compliance are part of the package. Finance, healthcare, legal. Even if an AI generated the perfect software for you, the question remains: who takes responsibility? Who certifies? Who answers when something goes wrong?
Finally, maybe, ultra-high-end products with exceptional UX. Some experiences are crafted in a way that isn’t just “it works,” it’s perceived quality. Notion, Linear, Figma. Replicating features is one thing. Replicating that coherence, that detail, that aesthetic and operational trust is another.
What risks suffering the most, instead, are niche utilities. The apps that “do one thing,” the $9.99-a-month micro-saas that live because they solve a specific problem better than others. If I can build it in an hour, that price starts to shake.
The barrier that remains, for now #
There’s a legitimate objection, and I feel it too.
People don’t want to build their own things. They want to use them.
It’s true. And it probably will be for quite a while. There’s convenience, there’s trust, there’s cognitive savings. Even just clearly formulating what you want is effort—let alone iterating, correcting, choosing.
But that barrier isn’t stable. It drops with habit. It drops with better interfaces. It drops with digital education. And it drops above all when the advantage of having something custom becomes obvious enough to justify the effort, while the effort keeps shrinking.
I think the inflection point won’t be “the tool has become accessible.” It will be “the first time someone I know shows me what they made in 20 minutes and I think: I could do that too.”
That moment is coming. For some people it’s already here.
What we might see in the next few years #
I’m not an analyst and I don’t have a crystal ball, so take these as informed feelings, not predictions.
I think we’ll see a first wave of consumer products that are no longer bought but built. At first in the most tech-friendly segments—developers, designers, university students—and then in concentric circles.
I think we’ll see enormous pressure on micro-saas pricing. Not because they become useless, but because it becomes hard to justify a subscription when the same utility can be generated on demand. Those who survive will offer something that isn’t easily generated: data, network, trust, deep integration with existing ecosystems.
And I think new intermediaries will emerge. Not “app builders,” but curators and distributors of software intentions. People who package prompts, configurations, already-tested flows, so others can create without starting from zero. A kind of template marketplace, but deeper, more operational.
There will probably also be attempts at regulation, more or less clumsy, around certain specific scenarios.
Breakfast as a metaphor #
I’ll go back to the starting point, because that’s where the feeling stuck to me.
It wasn’t the speed that really struck me. It was the cognitive cost. I was distracted, doing something else, with my head on two things in parallel. It wasn’t a work session. It wasn’t “today I’m building a product.”
And yet the result is good enough to be online.
For me, this is the signal. The fact that producing software is becoming something you do in parallel with something else, like sending an email or searching something on Google. An action that no longer requires a dedicated context, a specific skill, a special mental energy.
When something becomes like that, it changes its place in the cognitive and cultural hierarchy. And with it, the market that grew around it changes.
Software is becoming an intention. Not yet for everyone, not yet in a stable way, but that’s the direction, and the speed is increasing.
And so the question I’m left with, this morning, is this.
If you’re building a consumer product, is the value in the technological distance between you and your user, or in something that exists even when that distance no longer exists?
If you don’t have a clear answer, maybe it’s time to find one.