The Tool Won’t Change You. The Moment Will

I sat in a meeting this week and watched someone walk me through their process. Five systems. Manual touchpoints at every handover. Data re-entered at each stage. Silos between all of them. Nobody designed it to be this way, it just accumulated, layer by layer, over years. And now it’s just “how we do it”.

Same organisation. Same week. A manager used Claude to convert a manual spreadsheet into the exact import format their system required. Two clicks. Done. A task that had probably been a recurring pain for years, solved in an afternoon. Two realities. One company. Both completely normal in 2026.

This isn’t a technology problem

I want to be clear about something before we go any further. The people still running the five-system process, implemented a survival strategy that worked so they stuck with it. They built a way of getting through their day with the tools and knowledge they had. That’s not a character flaw. That’s competence.

They’re human. And humans do the best they can with what they have.

Humans are genuinely, deeply, structurally not great at change. Not as a character flaw, as a feature. We default to the known path because the known path is safe, predictable, and doesn’t require us to feel incompetent while we figure something new out. The spreadsheet process that takes hours has one significant advantage over the Claude solution that takes ten minutes: everyone already knows how to do it.

This is the part that technology rollouts consistently get wrong. They treat survival strategies as problems to overcome rather than evidence of capable people doing the best they could with what they had.

The goal isn’t to dismantle what people built. It’s to show them something better, and give them enough of a reason to try it.

Our sales teams have access to AI. They mostly don’t use it. Developers have unlimited access to all the AI models and still choose their familiar workflow every time. This isn’t obstruction. These are people who built expertise and reliability on the tools and processes they know. Asking them to change isn’t asking them to do something better, it’s asking them to temporarily feel less capable and uncertain. That’s a significant ask, and it deserves to be treated as one.

This plays out across every function, every industry, every level of seniority. New tools become available. Habits never shift. Access never equals adoption.

So what actually creates change?

Not a mandate. Not a lunch-and-learn. Not a knowledge transfer. Not a change management framework or a company-wide AI strategy document.

A moment.

One specific, personal, visceral experience where someone realises that the thing they’ve been dreading, the task with the heavy mental load attached to it, the one that’s been sitting on their to-do list for three weeks because it’s not hard, just deeply unenjoyable, is just gone. Done. In a fraction of the time they expected.

I heard about exactly this recently. Someone new to AI watched a colleague cut a week-long task down to an hour on the first try. The excitement in the room was apparently electric. Not “this is interesting technology” excitement. “What else can I try?” excitement. That’s a completely different thing.

A colleague showed me something this week that truly brought me joy. They’d taken a 300,000-line csv file, the kind that crashes Excel and defeats most humans on sight, and turned it into a clean, interactive PowerBI-style useable dashboard. Single prompt. First try. They weren’t just presenting a dashboard. They were bursting with the excitement of someone who has just watched the ceiling lift on what they thought was possible. That’s not ‘interesting technology’ excitement. That’s ‘what else can we do?’ excitement. And those are completely different things.

That moment matters because it creates intrinsic motivation, the only kind that actually builds lasting habits. Nobody ever permanently changed how they work because their were told to. They changed because the new way felt better than the old way. Relief is a powerful teacher.

And here’s what I truly find exciting about this:

AI doesn’t just make tasks faster. It makes the tasks you’ve been avoiding suddenly approachable and even enjoyable.

The report you’ve been putting off. The data clean-up job that’s been on the backlog forever. The thing that isn’t really that hard, just tedious enough to keep getting bumped. When that weight lifts, people don’t just complete the task. They go looking for the next one.

Then it gets infectious

Here’s where it stops being an individual story and starts being an organisational one.

That person goes home and tells their partner. Not “I used a large language model today”, nobody says that. They say “you won’t believe what I solved today, remember that annoying thing I’ve been complaining about for months.” Their partner tries it on something at their job. A colleague overhears the conversation and asks “wait, how did you do that?”

It spreads sideways. Not through a formal rollout. Not top-down. Through kitchen table conversations and Slack messages and “let me show you what I built”. Word of mouth for capability. And it travels at a completely different speed than any training program ever could.

The organisations genuinely cracking AI adoption aren’t running better workshops. They’re creating conditions for those first moments to happen, and then getting out of the way while people talk about them.

The manager who solved the spreadsheet problem? They’re doing more for AI adoption in that organisation than any tool subscription ever will. Because now someone saw it work. Someone with credibility. On a real problem. With a visible result.

That’s the template.

What the numbers actually tell us

For those who need the macro picture alongside the human one:

• Anthropic writes 70-90% of its own code with AI.
• Claude Code writes 90% of itself.
• Spotify hasn’t written traditional code in 2026.
• Google and Microsoft have AI generating roughly 30% of new code.
• Meta’s Zuckerberg has predicted AI will handle half of their software development within a year.

These aren’t start-ups experimenting. These are the organisations building the tools we all use.

You.com’s 2026 AI Predictions put four things on the horizon that are already present tense:

• Developers of the future being mostly AI engineers — people directing and reviewing rather than purely writing code. The role isn’t disappearing. It’s changing shape. The 300,000-line spreadsheet story is this prediction already in motion outside of engineering entirely.
• Every knowledge worker becomes a manager of AI agents — directing, reviewing, and deciding rather than purely executing. The skill didn’t disappear. It moved up a level.
• At least three job categories will begin seeing displacement of people unwilling or unable to adapt, starting with software developers — think about it this way. A decade ago nobody could say “I don’t do email” and keep their job. Five years ago the same became true for video calls. “I prefer my old workflow” is heading the same way. The timeline is shorter than most people are comfortable admitting.
• “A specific kind of ‘we’re doing well enough’ inertia will take hold in pockets of every organisation” — this is the most honest prediction in the document. Because that inertia isn’t irrational. It’s people with perfectly good survival strategies seeing no urgent reason to update them. Yet.

I’m not a developer. I can’t write Python from scratch. I have a Claude Pro subscription and a habit of asking “could I build that?” and then actually trying. In the last year I’ve built automation scripts, a RAG agent, a transcription app, a website, and a workflow tool. None of it required me to change career. All of it required me to be willing to feel uncertain while I figured it out.

That willingness is the actual skill. And it’s learnable, but only after the moment.

The question for leaders

If you’re running a team, a function, or an organisation, the real question isn’t “have we given people access to AI tools?”

It’s this:

have we created conditions for people to have their moment?

That means making space for experimentation without judgment. It means leadership visibly solving real problems with AI, not presenting about AI strategy, actually using the tools where others can see it. It means celebrating the person who cut their weekly report from two hours to fifteen minutes, not just the person who shipped the biggest project.

Shopify made AI effectiveness a fundamental expectation of employment. That mandate only works if there’s genuine support behind it. Otherwise you get compliance theatre, people who say they use AI and don’t.

Habits don’t break through pressure. They break through experience. The moment someone feels the relief of a task they’ve been dreading just disappear, that’s when the old habit loses its grip.

Your job as a leader is to make that moment more likely. Create permission to experiment. Model it visibly. Celebrate the small wins loudly. The rest takes care of itself.

The technology is almost incidental. The human experience of it is everything.