Right now, thousands of companies are building products and vibe coding half the features on their roadmaps.
Those products aren’t necessarily badly built. The UI is clean, the layouts and views are good, the integrations work. Someone spent real time making it all run, and keeps the build stable, bug-free, and fast.
But most of them share the same problem: nobody keeps using them after the trial.
The teams believed users needed every feature they shipped. What users actually got was too many options and too much flexibility, when all they wanted was to get one thing done. The tool solved the wrong problem, and solved it precisely.
That isn’t a new failure. It’s an old one, running at a new speed.
We can build faster than ever, but very few teams are trying to learn as much from each failure as they used to.
Fine is the new floor
You can vibe-code a product that works now. It’ll be fine: decently built, reasonable to look at, nothing obviously broken.
Anton Sten put words to what happens next in Designing for Trust: somewhere in it, you’ll hesitate, in a way you can’t quite point to.
That sentence describes most of what I see shipping right now, products that function and still feel slightly off. In Taste Cannot Be Delegated, Aurélie names shortcomings of an AI-generated outcome even more bluntly:
“Nothing is particularly memorable either. That is what design by committee usually produces: not failure, forgettability.”
I keep finding the same three things:
- Taste: knowing what belongs and what doesn’t
- Thought: the moment that makes it feel like someone cared
- Trust: the absence of that hesitation
None of the three comes from the model. All three come from the same place: time spent close to the people you’re building for. Researching the steps that teams would skip first when building has gotten this cheap.
Also in September: Taste Is Judgment, Not Preference · Trust Is the One Nobody Names · The Four Cs Before the Prompt
Daniel Arevalo is a Product Architect and strategic product design leader. He helps product teams turn AI capability into trusted, adoptable workflows, with a background in industrial AI, IT/OT cybersecurity, and high-stakes enterprise software for critical infrastructure.