Are Web Apps getting more or less important?

AI can do more than ever. Building a web app is probably one of the most visible examples: describe what you want, let an agent write the code, and you can have something running surprisingly quickly.

That makes me wonder how much time we, as developers, should still spend specializing in this kind of work.

At the same time, more and more software is moving into the browser. Web apps seem to be getting more important, while building them becomes less of a special skill.

Both can be true.

The same app, again

Think about the structure of a typical web app. A frontend, a backend, a database. Users log in, create something, edit it, and eventually delete it.

Of course the business requirements differ, but a lot of the implementation is still CRUD with a new, pretty frontend.

These are exactly the kinds of tasks LLMs tend to handle well. The same patterns have been implemented and documented countless times before. Another form, another API endpoint, another table of results.

As a developer, I would be careful about making this repetitive work the main thing I specialize in.

It also becomes harder to distinguish good and bad developers by looking at the result. A nice interface and a working demo tell you very little about how well someone understands the system behind it.

There is still a difference, obviously. It shows up when the requirements change, permissions get complicated, or the app starts losing data. But you might not see any of that in the demo.

I guess you could call this a lower skill ceiling for the standard CRUD app. Once an LLM can do most of the familiar work, getting even faster at that work gives you less of an advantage. The difficult parts of software engineering are still there.

More software in the browser

Meanwhile, the browser keeps taking over work that used to require a desktop application. Even Photoshop has a web version.

For users, the appeal is easy to understand. Open a link, log in, and get started. For companies, deploying updates centrally and supporting users across operating systems is attractive as well.

So the platform itself still matters. Making it easier to build web apps might actually mean we end up with even more of them.

The RAM problem

Then there is the current RAM crisis. Micron expects memory supply to remain constrained through 2027 and 2028 (source).

My expectation is that expensive local hardware gives companies another reason to move demanding work onto shared servers. Users can access the result through a browser without upgrading every machine to run the full workload locally.

Servers need RAM too, obviously. And putting an app in the browser does not automatically reduce its memory usage. The benefit comes when the expensive computation happens remotely and the hardware can serve many users.

For that kind of application, I think the pressure to move compute off the user’s device will grow.

Keeping business logic on the server

There is another reason companies might prefer this setup: protecting how their software works.

If you ship a binary to someone’s computer, they can inspect it. Reverse engineering has always been possible, and LLMs can help make sense of compiled code. Research such as BinMetric shows useful capabilities, although reliably reconstructing arbitrary software is still a much harder problem.

For a company whose advantage depends on a particular algorithm or business logic, that matters. Compiling the code was never a guarantee that nobody could understand it. As the tools improve, relying on that difficulty becomes less convincing.

Running the valuable logic on your own servers means users never receive that implementation. They can still study the inputs and outputs, but they have less to work with.

The frontend remains visible. The knowledge you want to protect stays on the backend. I think that gives companies another incentive to deliver their software as a web app.

What does this mean for us?

Web apps can become more important while routine web development becomes less valuable as a specialization.

For me, that means looking beyond the next framework or another way to build the same dashboard. Understanding the business problem, deciding where code should run, and knowing why a system fails are skills worth spending time on.

As I’ve written before, we still need to understand the software we build, even when an LLM writes parts of it.

Keep building web apps. Just make sure you are still learning something along the way.