We don’t just buy software. We buy the team, the years of testing, the documentation, the support, and the accountability. All of that is baked in before we even open the app. AI can replicate the output. It cannot replicate any of that, and I think we need to start talking about what that actually means.

In previous blog posts, I have written about my frustrations with large software companies: price hikes, captive markets, and creeping monopolies. But none of that changes what we stand to lose when we start replacing professional tools with AI-generated alternatives.

I want to reiterate: I am not anti-AI. I understand their value, and I genuinely believe they have a place in how we work and learn. But I am deeply concerned about how freely and uncritically they are being applied. This topic came to my mind again after watching a recent video. It showed someone who had essentially built their own eLearning authoring software using AI. Think Articulate Rise, but homemade. They then used that software to produce HTML-based learning experiences, again broadly similar to what a professional tool like Rise would output. It was impressive. It was also, I think, a problem!

The question is not whether AI can do something. It is whether it should.

When we use established software like Articulate Rise, we are not just buying a tool. We are buying into a whole ecosystem of quality assurance, testing, documentation, ongoing maintenance, and support. There is a company behind it whose entire purpose is to make sure that software is reliable, accessible, and fit for purpose. They manage the updates. They fix the bugs. They write the documentation. They are accountable.

Now imagine a world where everyone starts building their own versions of these tools. Who manages and maintains them? Who tests them properly across browsers, devices, and assistive technologies? What happens when something breaks mid-course, mid-launch, or mid-client-contract? Where is the documentation? Where is the support?

Don’t get me wrong, I create my own software. I have a few apps I’ve built to support my own workflows and honestly, I love doing it. But there is a big difference between building something for yourself and handing a Frankenstein solution to a client as if it were a professional, tested, reliable product.

This is not a hypothetical. This is where we’ll be if we don’t have more serious conversations about due diligence. We trust software companies for a reason. They provide something that goes far beyond the code itself: repeatability, reliability, and responsibility. That trust has been earned through years of iteration, user feedback, and professional standards. You cannot replicate that overnight with a prompt.

But this is just the cherry on the top. I’m starting to see more and more people using AI agents with disastrous consequences. The broader concern I keep returning to is this: what will it actually take for meaningful AI regulation to arrive? At the moment, the tools are moving faster than the frameworks designed to manage them. And in professional fields where the quality and integrity of what we produce genuinely matters, that gap is worrying.

The conversation needs to move on! Less about what AI can build, and more about what we should actually be building with it.