I saw a statistic the other day that stopped me mid-scroll.

It said that 78% of organisations have adopted AI in some form, but 74% report they’ve seen “minimal to no business impact” from their investment.

Now, I’m not a mathematician, but even I can tell you that those two numbers don’t live in the same house comfortably. If nearly four out of five organisations are doing the thing, and nearly three out of four aren’t getting value from it, then either the definition of “adoption” is doing some heavy lifting, or something fundamental is broken in how we’re approaching this technology.

I suspect it’s both.

What “Adoption” Actually Means

Let’s be honest with ourselves for a moment. When a vendor survey asks “have you adopted AI?”, what they’re really measuring is:

• Someone in the organisation has logged into ChatGPT
• A team bought a licence for Copilot
• IT deployed a chatbot that redirects to the same FAQ page the old system did
• The CTO mentioned “AI” in an all-hands meeting

None of these are adoption. They’re exposure. They’re the equivalent of saying you’ve “adopted running” because you bought a pair of shoes and walked to the letterbox once.

Real adoption means the technology is embedded in a workflow. It’s changing how a decision is made, how a process runs, or how a customer is served. It’s not a toy. It’s a tool.

And tools need to be fitted to the hand that’s using them.

The Hammer Problem

There’s a pattern I see repeating across the enterprise landscape, and it’s not new. It happened with cloud, it happened with agile, it happened with blockchain, and now it’s happening with AI.

Someone senior reads an article. They get excited. They mandate adoption. Teams scramble to bolt the technology onto existing processes without stopping to ask the most important question:

What problem are we actually solving?

The result is a lot of expensive hammers looking for nails. And when the hammer doesn’t magically fix everything, the organisation declares the technology overhyped and moves on to the next thing.

But the technology wasn’t the problem. The approach was.

Where the Value Actually Lives

I’ve been building small AI systems – nothing enterprise-scale, just personal projects and experiments – and the pattern I’ve noticed is this: the value doesn’t come from the model. It comes from the harness around it.

The model is just a next-token predictor. It doesn’t know it’s solving a business problem. It doesn’t care about your KPIs. It’s generating text (or code, or analysis) based on patterns in its training data.

The value comes from:

• The prompt – how you frame the task
• The context – what information you give it to work with
• The guardrails – how you validate its output before it touches anything real
• The integration – how it connects to the systems and workflows that actually matter

If you skip any of these four, you’re not adopting AI. You’re just adding a stochastic parrot to your team and hoping it doesn’t swear at a customer.

A Better Way to Think About It

Here’s a framing I’ve been using in my own work.

Instead of asking “how do we add AI to this process?”, ask “what part of this process is currently bottlenecked by human judgement that follows a pattern?”

If the answer is “none of it”, AI probably isn’t your solution.

If the answer is “this specific step where someone reads a report and makes a decision based on known criteria”, then you have a candidate. Not for replacing the person – for augmenting them. Give the model the grunt work. Let the human do the actual thinking.

That’s where the 74% who aren’t seeing value are getting it wrong. They’re trying to replace the thinking instead of the grunt work.

The Bottom Line

The 78% adoption number is probably real. The 74% disappointment number is probably real too. And the gap between them isn’t a technology problem – it’s a strategy problem.

We’re treating AI like a magic wand when we should be treating it like a power tool. A power tool in the hands of someone who doesn’t know what they’re building is just a way to make a mess faster.

But a power tool in the hands of someone who knows exactly what they’re building?

That’s where the math starts to work.