My next competitor is a prompt

A customer used to say, “I want a website, but I don't know how to build one.”

Today they can open an AI tool, describe the idea and get an initial interface.

Sometimes it looks pretty good.

If you run a small digital agency, that raises an uncomfortable question:

If a prompt can generate a website, why would somebody pay me?

I could say AI doesn't do “real development.”

That would be too easy.

It already does useful work. Ignoring that won't make it disappear.

What's becoming easier

Mockups, landing pages, simple data structures and prototypes are more accessible than a few years ago.

For someone testing a business idea, that's good news.

They can see a first version, change the design and decide what to build before spending heavily.

That lowers barriers.

An agency that charges only for producing ordinary screens may find it increasingly hard to explain its value.

The problem doesn't disappear with the code

Imagine a company that wants to manage visits by its technicians.

It doesn't really want “an application.”

It wants to know who is working where, how long visits take, what's still pending and how clients are informed.

Designing the right tool means understanding procedures, constraints and exceptions nobody has fully written down.

AI can generate components and functions.

But asking the right questions before building remains essential.

And deployment responsibility doesn't disappear because the code was generated.

What I need to sell better

As I try to grow KAMIT, I don't want to pretend every handwritten line is better than AI-generated code.

I'd rather focus on what actually helps a client:

• Understanding needs rather than blindly following a feature list.
• Designing a simple user journey.
• Defining business rules and edge cases.
• Protecting access and data.
• Delivering tested, documented, maintainable software.
• Supporting the product as it changes.

These aren't just technical tasks.

They're commitments to an outcome.

Will prices always go down?

It's tempting to think so.

When some steps become faster, their costs may fall.

But a project isn't just the hours spent typing code.

Scoping, integrations, quality, security and support still exist.

Pricing may need to reflect value and responsibility more clearly than the amount of code produced.

I explore related questions in The tiny SaaS dream.

What about beginners?

I don't think the right message is, “AI will do everything for you.”

That would skip too much learning.

You still need to understand data, networks, errors and security to evaluate the tools' output.

Generated code isn't automatically correct.

It's material you need the skills to review.

Where I want to fit

My next competitor could be a prompt.

Or a developer who uses AI better than I do.

Or a small team that deeply understands a particular business problem.

That isn't necessarily bad news. It forces me to clarify what I provide.

If my only argument is “I can write code,” I'll need a better one.

If I can help turn real problems into reliable services, there will still be room for me.

The tools will change.

The need to understand won't.