This article was originally written in June of 2024. A lot has changed since then, especially with the rise of LLMs, so it felt worth revisiting.
I came across an article titled, “I have never used a library, framework, or an API. I don’t know what to do.” The title alone tells you everything you need to know. Whether the article itself holds up is beside the point. The concern is real.
I graduated from Austin Community College in May 2024 with an AAS in Web Programming. Looking back, I can see how someone could come out of a program like that and still feel unprepared to actually build something in the real world.
The path was straightforward. HTML and CSS, then JavaScript, then a brief introduction to frameworks and APIs. And when I say “brief,” I mean exactly that. We were expected to pick things up quickly and fill in the gaps on our own. That part, at least, turned out to be accurate to how the industry works.
I worked with PHP and CodeIgniter, built and deployed a project, and even integrated a basic API. It was enough to get exposure, but not enough to build confidence. The real learning started after graduation.
Like most developers at that stage, I started chasing frameworks. Flask, Django, Laravel, React. I spent more time deciding what to learn than actually building anything. Eventually, I forced myself to pick one and move forward. That decision mattered more than the framework itself.
I chose Flask and built out a site. Then I ran into the realities that tutorials do not cover. Hosting. Security. Server configuration. I set up my own server, got it running, and then had to deal with traffic and attacks I was not prepared for. It worked, but it made one thing clear very quickly. Just because you can run something yourself does not mean you should.
That experience pushed me toward a different direction. I started looking for solutions that reduced the amount of infrastructure I had to manage. That naturally led me toward frontend frameworks and eventually into React.
At that point, I made another decision that seemed logical at the time. Separate frontend and backend. Build everything myself. Learn everything at once.
Was that a mistake? In a way, yes.
I learned a lot from it. More than I would have in a controlled environment. And to be fair, tools like ChatGPT did not push back on that decision. It was more than willing to help me move forward, even when I was stacking complexity on top of complexity.
But I followed through. I built a CRM. It started as a learning project, but it did not stay that way. I now have a team of salesmen using it. It became a real product.
I am still working on it. It is still evolving. I am improving validation, tightening the architecture, and planning the next version, all while balancing client work.
So yes, it was a mistake, but not because the approach itself was wrong. It was a mistake because I introduced too many variables at once. Learning React, TypeScript, API design, authentication, and deployment all at the same time slowed everything down.
That experience is what forced my perspective to shift.
If I were starting over today, I would not split the stack immediately. I would start with something like Next.js.
Next.js removes a lot of the friction I ran into early on. Routing, API routes, server-side rendering, deployment. It brings those concerns into one place without forcing you to manage the entire infrastructure yourself.
More importantly, it lets you focus on building features instead of wiring systems together.
That does not mean you avoid learning the underlying pieces. It means you learn them in context, instead of all at once.
There is a difference between understanding how something works and trying to build every layer of it from scratch on day one.
This is where things get uncomfortable.
With tools like ChatGPT, you can spin up a Next.js app, generate components, wire up API routes, and even debug issues without fully understanding what is happening under the hood.
So the question becomes: if AI can do most of the heavy lifting, is there still value in learning frameworks at all?
I think there is, but the reason has changed.
Learning a framework is no longer about memorizing syntax or knowing every configuration option. AI can handle that. The value now is in understanding structure, tradeoffs, and how pieces fit together.
If something breaks, you still need to recognize why. If performance is an issue, you need to understand where to look. If a feature needs to scale, you need to know what decisions matter.
AI can generate code, but it does not carry responsibility for that code. You do.
That shifts the role of the developer from writing everything manually to making informed decisions about what is being built.
Looking back, the problem was never the lack of frameworks in school. It was the lack of context.
I learned concepts, but I did not yet understand how those concepts played out in real projects. That only came from building things, making mistakes, and adjusting.
Frameworks still matter. Next.js, React, and others are not going anywhere. But the way we approach them has changed.
You do not need to master everything before you start. You do not need to build every layer yourself to understand it. And you do not need to compete with AI at writing code line by line.
What you do need is the ability to recognize what you are building, why it works, and where it can fail.
That is the part that does not get automated.