Last night I was talking to a young senior project manager who works at a university, and I caught myself asking my usual question: if he had started using AI yet.

His response was positive, and I was happy to hear that the university has been encouraging everyone to try it and has given people Copilot licences. He has been using it a bit. Mostly for meeting notes. Then he said he still enjoys doing the critical thinking himself. That is the part of the job he likes.

I have heard some version of that answer a few times now, and I was glad he was actually using it.

Normally this is where I go into my standard explanation. That he is the subject matter expert. That AI does not replace his thinking, it helps him move faster. That it can act as a kind of critic — something to push back on his ideas, ask the awkward questions, stress-test what he already knows.

I know that explanation well. I have said it often enough!

But later, thinking about the conversation again, it struck me that the explanation felt small compared to what we are now seeing in AI tools.

Not wrong. Just incomplete.

Over the last few weeks, watching what is happening with tools like Claude Code, I have had a growing sense that using AI to refine writing, do research, or critique your work is useful, but it is not the real shift. The barrier to creating automation workflows, with inbuilt agentic AI, is now here for the masses. That feels structural. It feels like something underneath the work itself is moving.

I feel this is where the conversation should be, rather than focusing on getting people to use AI for writing.

A couple of years ago I worked with an automation company using one of the large enterprise platforms. Automating a business process was heavy work. Discovery. Specialists. Long timelines. Significant budgets. It was powerful but slow, and only certain organisations could afford to do it properly.

Automation felt like something you implemented.

What feels different now is how light and accessible it has become.

With tools like Claude Code, anyone in a business doing a repetitive task can build a small capability for themselves. Something that runs in the background. Something they monitor while doing other work. They do not stop working. They start supervising work instead.

That shift feels like a turning point.

If businesses are not testing how they can use this technology to evolve, they may find their business model is no longer viable in the future.

That is the part I keep coming back to.

Large enterprise organisations often cannot move quickly. Small to medium sized businesses, however, have a real opportunity. They are agile enough to change how they work. They are close enough to their processes to see what is actually happening.

At the same time, they still have silos. Small systems. Local workarounds. Processes that have grown inside teams and stayed because they mostly work.

That is where this becomes interesting.

It is not about automating everything. It is about looking across those processes and spotting the smaller ones, the repetitive or invisible parts, that could genuinely change how people work day to day.

Finding where automation frees people up instead of getting in the way.

And having someone inside the business who can hold that view. Someone who can look across teams, notice patterns, and put enough structure around change so it does not become chaotic.

The work shifts away from doing the process and toward shaping it. Watching it. Governing it. Adjusting it as it runs.

That is the part that feels exciting to me.

AI tools are moving so fast that what we build today may do much more in a few months, or be built directly into the platform itself. Changes need to be implemented lightly, in a way that allows them to evolve as the tooling evolves.

Every business needs a champion. Someone to educate. Provide governance and security. Help the organisation use these tools to evolve with them.

Business models as we know them will change.

It feels like the beginning of a different kind of work, and I cannot stop thinking about what that might look like as it unfolds.