Insights

We’ve heard of enterprise organisations building internal tools using AI-generated code, only to discover they are locked into fragile systems that nobody can properly explain or support. Once you’ve committed to a direction without governance, it’s hard to claw it back.

Data governance is often overlooked

There’s another layer that many businesses forget: governance.

When you use AI tools — especially SaaS-based ones — you’re often sending code, content, and sometimes sensitive or proprietary data to a third-party platform. That platform might store it, learn from it, or even make it available to other users depending on how the terms of service are written.

This isn’t just about technical risk. It’s about legal, compliance and reputational risk too.

Who has access to the prompts and code you’re generating? Where is that data stored? Is it encrypted? Is it being retained? Could it be used to train other AI models? In many cases, the answer isn’t clear...and in regulated industries, that’s a serious issue.

It’s easy to imagine a scenario where a well-meaning team member pastes sensitive logic or client data into an AI tool to “speed things up”, without realising they’ve just breached policy.

And yet, very few companies are asking these questions.

So before you start feeding your systems and your IP into a black box, make sure someone is asking: “Where is this data going, who has access to it, and how will we support the results?”

The lesson from in-house hiring

We made a similar argument in our article Should I hire an in-house developer?

There, we talked about the risks of relying on a single person to deliver something that really needs a team. Not because that person isn’t smart or skilled, but because systems development isn’t just about writing code. It’s about project management, user experience, accessibility, quality assurance, digital marketing, long-term scalability, and security.

AI doesn’t change that equation. If anything, it adds a new layer of risk. Because at least with a developer, you can ask what they did and why. With AI, you’re often left with a wall of code and no explanation.

So where does AI add value?

Used wisely, AI is a powerful productivity booster.

A skilled developer might use it to scaffold components. A tester might use it to generate quick scripts. A strategist might feed it prompts to explore variations in user journeys or messaging.

In the right hands, AI can make a good team faster. What it cannot do is replace the judgment, experience, and human thinking that underpins high quality software and digital platforms.

When you use AI as a team member, not a team replacement, you benefit from the best of both worlds. Faster output, paired with expert oversight. Innovation, paired with responsibility. Speed, without giving up control.

If you’re building something that matters

If you’re building a business-critical website, platform or application, then AI should be part of your toolkit. But only part.

Relying on it fully is like building a house without an architect or surveyor. It might go up fast. It might even look the part. But one day the wiring will fail, the pipes will burst, or the foundations will shift. And when they do, there’s no one to call who knows how it was built.

That’s not efficiency. That’s fragility.

So yes, explore what AI can do. Use it. Learn from it. Build with it.

But build with people too.

Because tools don't carry risk. People do. And the right team will help you move fast without breaking everything.