Lovable vs traditional web developer

There’s a lot of debate around this.

From my experience, AI significantly reduces time to production—projects that once took months or weeks can now be completed in days or hours. Refactoring is faster as well.

That said, when AI breaks down, a developer still has to understand and fix the code. I’ve heard of teams abandoning AI-generated projects because no one fully understood the codebase.

Bottom line: AI is a productivity tool, not a replacement. It can help, but it’s not a silver bullet.

Marketing is still marketing—SEO, targeting, and messaging matter. Relying entirely on AI for marketing content can be risky if it gets things wrong.

Used carefully, AI is valuable. Used blindly, it can create problems.

Think of AI as a way to do what you were going to do anyway but quicker.

As AI can write code quicker, it can also ruin codebases quicker too. You can protect yourself from this by ensuring you support your code with functional tests, and also ensure you have a good version control strategy (ie learn git, commit small and often).

It's totally possible to follow test driven development strategy working with Agentic code tools. You need to stay focussed, keep a check on the outputs of your methods and continue to review your own code before you publish to production.

You should also look at writing an agents.md file to describe to your AI how your code base is structured and how you prefer to write code. My agents.md describes how my commit messages are made, how to run linting and what conventions our code base is following. This is essentially a glorified readme.md specifically targeted at AI. Be careful not to be too verbose in here though as this file is passed to your AI with every prompt so will chew through the credits.

Web developers have a great future ahead of them if they remember that they are solving problems with code, not just there to churn out code. For me it's like I've gone up a step and instead of writing code, I spend more time on the acceptance criteria, testing and planning.

Back when I started in web, we had to make html by hand. Now AI's are enabling speeding through this kind of work and now I'm able to concentrate on the quality of output and to be able to prototype quicker (and be less precious of throwing stuff away).

I can't wait to see what we're going to get next!!

If programmers would just use it as a tool. I reworked a python tube bending program with AI just to see what it would do.

My original I did all the math (matrices) the old school way multiplying them out. AI came back with Numpy results like:

n1 = np.cross(v1, v2)

But in the background Numpy is still doing the same math. But the key thing is I understand the math and what the Numpy library is doing. So I suggest only use AI to help. But don't use AI if you couldn't program it without AI.

Here is the thing: AI could give code that works. But is it secure, is it dealing with user data correctly. The mis-use of AI can be a bad thing.