Building with AI: the plane got faster, the airport didn’t


AI made writing code much faster, but deciding what to build and finding out if anyone wants it is as slow as ever. Speed up only the building and you get more features nobody uses. The time AI saves is better spent somewhere else.
Engineers can now build in an afternoon what used to take a sprint. Founders scroll past posts about someone shipping a whole app over a weekend, and a worry sets in: if building is this easy, why aren't we doing more of it?
So teams speed up. Requests that sat in the backlog for months get built the same week. Asking "do we know anyone needs this?" starts to sound like an excuse, and nobody wants to be the person slowing everyone down.
I understand the fear. It's also expensive, because it confuses two speeds: how fast you can build something, and how fast you can find out whether it's worth building. AI changed the first one a lot. The second has barely moved.
Think of AI as a much faster plane: Brussels to New York in two hours instead of eight. You still have to get to the airport, check in, queue at security, wait at the gate and board. And before any of that, someone has to decide where you're going and why.
When the flight took eight hours, the airport was a small part of the trip. Cut the flight to two and the airport becomes most of it. Software works the same way. Writing the code got faster; the work around it didn't.
Writing the code got faster; the work around it didn’t.
Someone still has to decide which customer problem is worth solving this quarter. Someone still has to talk to customers and find out whether the problem is real, and whether they'd pay to have it solved. Product, engineering and sales still have to agree on it. After it ships, sales has to learn how to pitch it, customers have to find it, and the team has to support it for as long as it exists.
Cursor doesn't do any of that for you. If anything there's more of it now, because more gets shipped, and every shipped feature has to be explained and supported.
So when someone says AI made their team five times faster, it's fair to ask which part got faster. Usually it's the flight.
Pendo looked at usage data from 615 software products and found that 80% of features are rarely or never used [1]. That report came out in 2019, before AI coding tools. Back when building was slow and expensive, most of what teams built still went unused.
Make building five times faster, keep deciding what to build the same way, and you'll end up with five times as many features nobody uses.
The DORA research team surveys engineering teams every year. In their 2025 report on AI-assisted development they call AI an amplifier: it makes strong teams stronger and makes the problems of struggling teams bigger [2]. Nearly 90% of the people they surveyed (about 5,000) use AI at work. Delivery speed went up with AI use, and so did delivery instability.
Make building 5x faster, keep deciding what to build the same way, and you'll end up with 5x as many features nobody uses.
The cost of "we can build it, so let's build it" shows up months after the release:
The product gets harder to sell. Every extra feature is one more thing to explain in a demo and to new users.
The team ships a lot while revenue stays flat. That's tiring, and people usually feel it before the dashboards show it.
The customers who pay the most wait longer, because the team's attention went to whoever asked loudest.
All of it has to be maintained: bugs, support tickets and edge cases, for as long as the feature exists.
So the fear of looking slow pushes teams to speed up the part that was already fast, while the expensive part (deciding what to build) stays as slow as it ever was.
A week spent checking whether people want something costs far less than years of supporting a feature they don’t use.
You can't skip the airport, but you can get through it faster. This is where I'd start:
Ask who pays for it before asking how fast you can build it. If nobody can name a customer who'd pay more, stay longer or buy because of it, it waits.
Use the speed for discovery too. AI makes it cheap to build a rough version, show it to five customers and throw it away.
Write down what you've decided not to build and share it with the team. It takes pressure off them, and off you.
Check whether shipped features get used. A feature nobody opens three months after release is just cost.
Give someone the job of saying no. In most small teams nobody has it, so everything gets built.
As for looking slow: customers notice the releases that change something for them, and they don't count the rest. A week spent checking whether people want something costs far less than years of supporting a feature they don't use.
Sources
Pendo (2019). The 2019 feature adoption report.
Saunders, M., InfoQ (2025). DORA report finds AI is an amplifier in software development, but trust remains low.

Edita Dermontaité
Hands-on CPO and product leader who still builds (with AI)
I’ve often been the first: the first product hire at Leadcamp and Bizzy, where I led product as CPO for five years, and the first user researcher at Lighthouse, where I spent two years building the research function.
For the past year I’ve worked as a fractional CPO and hands-on product leader, helping B2B SaaS teams between traction and scale connect what engineering builds to what brings in revenue.
I started out as a graphic designer, so I still design and test things myself, these days mostly with AI: in Claude and Lovable, or straight in code with Cursor.
Book a free 30-minute call, or a coffee in Ghent. We’ll look at what’s slowing your product down and what to do about it. I work with teams fractionally or hands-on as an embedded product leader, 2–5 days a week, depending on your needs and my availability.
Product leadership for B2B SaaS between traction and scale, from a hands-on CPO and product leader who still builds (with AI). I take product decisions off the founder’s shoulders and get product, engineering and sales moving in one direction: revenue.

New North BV
BE 1025.547.940
9050 Ghent
Belgium