There is a feeling I encounter more and more when looking at new apps.
It is not necessarily 'this is good.' Nor is it necessarily 'AI is really scary.'
It feels more like a whisper: what I am seeing here is no longer as difficult to build as before.
That feeling does not come naturally. OpenAI has integrated apps right into ChatGPT. GitHub Spark straightforwardly describes that users can speak in natural language to receive a web application and deploy it with less friction than before. Replit is also heading in the same direction: from verbal descriptions to functional apps or websites.
The point worth keeping from this change does not lie in whether AI has completely replaced the product team or not. It lies elsewhere: the part of the app that users see and touch first is becoming less scarce.
In the past, simply having a functioning app, a sufficiently polished interface, and a smooth user flow was a fairly strong signal to outsiders. It does not prove the entire value of a project. But at least, it shows that the project has overcome a certain level of friction to turn the idea into something usable. When that friction layer decreases, that signal weakens as well. The practical value of the app may not decrease at the same speed. But its signaling value, meaning the ability to make others believe there is something worth the market price here, is likely to diminish first.
This is where the story touches on crypto.
This way of reading is not for every project. It is more accurate for a narrower group: projects where the valuation is still being supported significantly by looking at the product and concluding that there must be a strong core value behind it, while that core has not yet clearly emerged in liquidity, distribution, data, accumulated trust, or the right to access cash flow.
For that group, the app once did two things at once. It was the interface for users. But it was also surface evidence that there is something worth the market price here. When AI makes that surface evidence layer cheaper faster, the question for tokens also becomes harder: what is this token anchored to that is harder to replicate than the app that users are seeing?
That is when the paradox begins to emerge.
AI can indeed help a project move faster. It shortens the path from description to a functional product. It helps test faster, fix faster, and ship a basic app layer faster. But this also weakens a fairly shallow type of moat: a moat based on 'we have an app, we have shipped it.' As this spreads out more, the act of shipping itself is no longer a strong proof as it once was. It is still an achievement. It just is not rare in the same old sense.
From there, an old boundary began to become clearer.
On one side is the project using the app as the gateway to a truly scarce thing behind it. That thing could be liquidity, distribution, accumulated trust over time, or the right to access cash flow. For this type, AI may not necessarily weaken the investment thesis. In some cases, it even helps improve the interface and makes the operational surface more efficient.
The other side is the project using the app as the presentation layer of the narrative. Users see a product, so it is easy to believe that behind it there is a value system that is hard to replicate. But if that hard-to-replicate part is not clear, then when the cost of building the presentation layer decreases significantly, the gap between the surface and the core also reveals itself faster.
AI does not weaken all apps equally. It is weakening some uses of apps faster, such as through valuation evidence.
Therefore, perhaps a more reasonable reading is not 'AI will kill crypto projects with apps.' It is closer to this: AI is pushing the valuation process back to earlier in projects where the visible product carries too many signaling roles.
If the surface app layer becomes easier to make, the token will struggle to survive solely on the feeling of 'we have a product.' It will be questioned more rigorously about what lies behind: what liquidity, what network, what economic rights, or what distribution advantage can others not replicate just by a sufficiently beautiful app.
The market may still read in the old way for some time. But the pressure has begun to change direction.
And when it changes enough, the question is no longer just whether this project has an app or not.
The harder question will be: if this app layer can now also be built quite quickly, what remains behind it that is still scarce enough to support the value of the token?