I’ve been noticing something over the last month that’s been making me increasingly uncomfortable, so I wanted to share that in this month’s Build.

Let me give you an example that stuck in my mind. Last Tuesday, I was reading a Slack message from a founder I’m working with. It was perfectly constructed, well formatted with clear bullet points and a nice professional tone.

But still it took me reading it three times to realise what was really nagging at me: there was none of them in it. Just the ghost of a ChatGPT prompt. Nothing human in any way.

That got me thinking about what’s happening across many of the businesses I have the privilege of seeing inside right now.

The rise of management by AI

Over the past few months – particularly since commonly available AI tooling became properly sophisticated – I’ve noticed something shifting.

Not in what AI can do (we all know that’s transformative), but in what people think they should be doing with it.

You’ve probably seen it. You get reports that look accurate. Presentations that tick every box. Emails that sound… fine.

But they’re all surface, no depth. To me they’re the business equivalent of those AI-generated hands with seven fingers – everything looks right until you actually look at it.

I’ve written about AI workslop before.

But here’s what’s really concerning me: I’m increasingly seeing people use AI to do their actual managing.

That might be feedback delivered via Slack that’s obviously been produced by an Openclaw agent. It might be strategic direction issued by email that reads like it came from a prompt library. Sometimes it’s performance reviews that could have been written about literally anyone.

And look, I do get it. We’re all knackered and pushing hard to scale businesses. We’re all trying to do more with less. The temptation to let AI handle the difficult conversation or write the tricky email is enormous.

But this is also where it falls apart for founders.

When the questions start

I had another experience this month where I watched this play out.

A COO I know had used AI to provide feedback on a project deliverable. The message was comprehensive, technically accurate and well structured. So far so good you might think.

The team member has read it, had some questions about a particular point – a subtle, organisation-specific nuance that the AI had missed – and asked to discuss it.

The COO couldn’t engage meaningfully in the discussion that ensued.

That wasn’t because they didn’t want to, but because they genuinely didn’t understand the subtleties of their own feedback. They’d outsourced the thinking entirely.

And in a moment suddenly their credibility as an expert in their own domain was completely undermined.

This is the thing about management that people are missing when they automate themselves too much: management is an irreducibly human act.

Collaborating with others. Learning to navigate conflict. Dealing with ambiguity and risk. Reading the room. Knowing when to push and when to ease off.

These aren’t administrative tasks you can automate away.

They’re the actual work of leadership.

So what should you be using AI for?

Let’s be clear: I might be hitting 50 this year but this isn’t a luddite rant about avoiding AI. That ship has very much sailed and I’m super excited about what I can do with technology now that wasn’t even in my thinking a year ago.

AI is transforming how we work, how we scale and how we build businesses. If you’re not using it, you’re operating with one hand tied behind your back.

The question isn’t whether to use AI. It’s what to use it for.

Here’s a working thesis that I’m exploring:

Use AI to supercharge your strengths. If you’re brilliant at strategic thinking but rubbish at first drafts, let AI help you get words on the page faster. If you’re great with people but hate documentation, use it to capture and structure your thinking.

Use AI to shore up your weaknesses. Struggle with data analysis? Let AI help you spot patterns. Terrible at grammar? Fine, let it polish your prose a bit.

But don’t use AI to replace yourself.

Don’t use it to handle the difficult conversations you’re avoiding.

Don’t use it to provide feedback you haven’t actually thought through.

Don’t use it to sound clever about things you don’t understand.

Because here’s the thing: as a founder, you have something AI fundamentally cannot replace.

You have skin in the game.

You have context that goes back years.

You have intuition built from pattern recognition across hundreds of decisions.

You have relationships with your team that are built on trust and shared experience.

You have a vision that exists somewhere between your lived experience and your imagination.

That’s not something you can prompt engineer or capture in a Markdown file in your personal operating system.

The uncomfortable bit

What I’m seeing – and deep down I think this is what’s really bothering me – is founders and leaders using AI to avoid doing the hard work of leadership.

Having the difficult conversation.

Sitting with the ambiguity.

Making the decision with incomplete information.

Showing up as a whole human being rather than a polished professional persona.

But surely that’s inherent the role of the founder?

The founder role isn’t to produce perfectly formatted feedback or comprehensive strategy documents. The role is to lead people through uncertainty, building trust and making others others better.

And you cannot outsource that to a large language model.

So here’s what I’m asking myself (and what I’d invite you to ask yourself):

• What am I using AI to avoid rather than to enable?
• Where am I substituting algorithm for authenticity?
• What conversations am I having with Claude that I should be having with actual humans?

Because the real risk isn’t that AI will replace us.

It’s that we’ll replace ourselves with worse versions of ourselves before it gets the chance.