I saw a post from Kiran Patil at Horizontal Digital recently that stuck with me.

His analogy was very simple: AI is the map. You’re still the driver.

Full credit to Kiran for the idea, because it keeps appearing in conversations I’m having about AI. The more I’ve thought about it, though, the more I think there’s another layer to it to extend what he wrote.

The conversation around AI is still overwhelmingly focused on what it can do.

• Can it write the code – Yes but whether it’s good code well that’s another story
• Create the content – Yes it can but everything can start feeling the same
• Analyse the data – Yes this is one of the best parts
• Design the architecture – Not really design but it can find flaws and check ideas
• Build the presentation – Yes to a point but same as content everything starts to look the same

Increasingly, the answer is yes. And that’s exactly why I think we’re approaching the wrong question.

The interesting question isn’t whether AI can get us from A to B. It’s whether B is the correct destination.

Think about how we used to travel before sat navs and mobile phones. You planned the journey with a map book, worked out the route and then tried to follow it.

I remember using Microsoft AutoRoute for skiing trips to the south of France. The journey would be printed across about 20 pages of paper. Miss a turning and things got interesting very quickly as you tried to work out where the hell you were and somehow pick up the route again.

When I first started driving, I had a map book in the car. Then I got a TomTom stuck to the windscreen with live traffic. Put in the postcode and follow the directions.

Now I can get into almost any car, connect Apple CarPlay and drive somewhere I’ve never been before without thinking much about the route at all. Technology progressively removed parts of the job.

AI is doing exactly the same thing to knowledge work.

I can describe an application and have AI generate a significant amount of the code. I can give it thousands of pages of content and find patterns that might once have taken a team weeks to identify. I can go from an idea to something tangible incredibly quickly.

That doesn’t mean I blindly trust the output, especially in production. But the amount of work between an idea and an outcome is shrinking dramatically.

The future is now

This isn’t something that’s coming. It’s already happening, and it’s only going to get better.

This is where I’d extend Kiran’s analogy. Today, I’d absolutely keep my hands on the wheel. But I don’t think we should build our AI strategy around the assumption that humans will always need to drive.

Agents will increasingly execute tasks and communicate with other agents. Within defined boundaries, they’ll build things, test them, deploy them, monitor the results and adjust based on what happens next.

Some workflows that currently require ten people might require two. Some might eventually require none. That doesn’t make the human irrelevant. It changes where the human provides value.

I’ve spent a lot of my career around CMS, DAM, PIM and digital experience platforms. For years, a big part of the expertise was knowing how to make these platforms do things: modelling content, connecting systems, building components, migrating assets and deploying applications.

Those skills still matter, but AI is rapidly reducing the cost of the how. Things that once required hours of research, specialist knowledge or development can increasingly be accelerated or handled by AI.

What AI isn’t making nearly as easy is the why. Why are we building this? Why does the customer actually need it? Why are we migrating thousands of assets that nobody uses?

As the how gets cheaper, understanding the why becomes more valuable. And perhaps the most valuable question is: should we be doing this at all?

That matters because AI can make an organisation incredibly efficient at doing the wrong thing.

A document that took four hours now takes twenty minutes. A developer builds something in a day instead of a week. A marketing team creates fifty pieces of content instead of five. That productivity is brilliant, but only if the document was needed, the application solves the right problem and those fifty pieces of content are worth creating.

Customers aren’t ultimately paying us to produce Jira tickets, architecture diagrams, PowerPoints or code. Those are outputs. They’re paying us to help them make better decisions and solve the right problems.

If AI gives us more time, that’s where I think we should spend it: understanding the business, challenging assumptions, connecting things that weren’t previously connected and deciding what should actually happen next.

The consultant who simply knows how to operate the technology will face increasing competition from AI.

The consultant who understands the technology and can identify which problem is worth solving becomes more valuable. The same applies to developers, architects, marketers and content teams.

AI doesn’t remove expertise. It exposes the difference between expertise and execution.

Today, AI is the map. Increasingly, it’ll become the sat nav. Then it’ll start driving. Eventually, for some journeys, we’ll probably just tell it where we want to go.

But somebody still needs to choose the destination, understand why we’re going there and recognise when that destination needs to change.

That’s where I think human value is moving.

So yes, keep your hands on the wheel today. But don’t build your career around being better at turning the steering wheel.

Get really good at knowing where we should go next.