Hey friends,

This week I’ve been thinking about a mistake a lot of leaders make with AI. They treat it like an IT rollout. New tool. New policy. New training deck. Then a tidy “go-live” date.

And in most organisations, that approach feels safe… right up until you realise something uncomfortable:

By the time you finish the rollout, the world has already moved on.

That was the thread running through my conversation with Wayne Pales, General Manager of Digital Business at Barwon Water. Wayne’s been leading large-scale transformations for 20+ years across energy and water (AGL, AEMO, China Light Power), and what stood out wasn’t just what they’re building — it’s how they’re getting people to move with it.

Because this isn’t an IT thing. It’s a human thing.

Wayne put it simply: the moment people hear “technology”, a chunk of the workforce silently steps back.

“I’m not good with technology." That sentence kills momentum before it starts. But if you reframe AI as a conversation, the door opens.

If you can explain a problem clearly, like you would to a colleague or a consultant, you can work with AI. That’s what lowers the barrier. Not more features. Not more licenses.

A different mental model.

Here’s the move I loved:

Instead of writing an AI strategy upfront, Wayne and his team ran a three-month program with the executive team, senior leaders, and the people team.

Every week: 90 minutes. Hands-on prompting → deeper capability → eventually some leaders building basic agents. And in between: ethics, governance, risk, leadership conversations.

The goal wasn’t to “train them” in a classroom way. It was to make leaders confident enough to lead adoption without outsourcing their understanding to IT.

Because Wayne’s point is dead right:

How can the business help shape strategy if they don’t understand what’s possible?

Wayne described four areas they’re focusing on. What I like is how practical they are, no hype, no “AGI will solve everything”, just foundations that scale.

1) Platform (the tech foundation)

They’ve done the considered part: controlled tools, governance, and guardrails. Not a free-for-all.

2) Data (the fuel)

Wayne said something that every organisation needs to hear:

If AI is fuelled by data, you need a culture that’s obsessed with data quality.

Not in a perfectionist way. In a “we care enough to notice when it’s wrong” way.

They’re even experimenting with agents that can detect data integrity issues and surface them to subject matter experts to clean up. That’s smart because instead of waiting for perfect data, you use AI to help you get there.

3) Process (because agents will automate whatever mess you already have)

This was one of the strongest moments.

As they lean into agents, they’re forcing the business to articulate workflows that used to live in people’s heads. The tacit knowledge.

Instead of asking teams to draw formal process maps (which most people hate), they’re recording conversations and using AI to draft process models, then going back to the experts to validate:

“Is this what you actually do?”

That’s how you find automation opportunities without turning it into a bureaucratic exercise.

And Wayne’s warning matters:

If you don’t improve the process first, you’ll just get agents performing bad processes faster.

4) People (turning adoption into movement)

This is the part most organisations underestimate. Different people are at different stages:

So Barwon Water is leaning into a “movement” model: support the people who are keen, and let curiosity spread through peer influence over time. Which brings me to the part that surprised me most…

Wayne’s team started with a 50-seat trial and then rolled Microsoft Copilot out to roughly 450–500 people.

Not because they had a neat ROI spreadsheet upfront.

But because they wanted to create an environment where experimentation could actually happen, while they build measurement frameworks in parallel.

The rule was simple:

If you want access, complete the compliance training. Learn the guardrails. Then go play.

And instead of “call the help desk”, they created an internal AI Hub, a social space for questions, prompts, and shared learnings. That’s an important detail, because it keeps AI from being framed as “IT support”.

It becomes business creativity. Business imagination.

This was one of the most practical answers of the whole episode.

Shadow AI is happening everywhere. The question is what you do about it.

Some organisations respond by locking everything down.

Wayne’s view (for now) is different:

Because if you lock it down too hard, people don’t stop. They just move to their personal devices and the risk gets worse. That’s a very grown-up way to look at it.

Barwon Water faced the same thing many teams face: leaders trained on ChatGPT, then moved to Copilot for governance reasons, then heard complaints.

Wayne said Microsoft supported them by showing equivalence, sometimes you just need to do it slightly differently. And the reality is the tools are moving so fast that what didn’t work a month ago might work brilliantly now.

Which is why they’re experimenting with something I think every organisation will eventually need:

Agents that surface role-specific updates.

Instead of expecting everyone to keep up with the firehose, you give finance teams finance-relevant Copilot updates, operations teams ops-relevant updates, and so on. Because “traditional change management” can’t keep up with monthly product shifts.

Wayne’s most grounded advice was also the most useful:

The people who say “AI hallucinates” often gave it vague input.

If you can clearly explain the problem, like you would to a smart human, the quality goes up dramatically. And if you can’t explain it? Start there.

Tell AI: “I don’t know how to articulate this. Ask me questions until the problem is clear.”

That’s the skill. Not prompt tricks. Problem clarity.

Wayne’s prediction for the next 3–5 years was clear:

We will have an agent workforce embedded inside teams. Procurement will have agents. HR will have agents. Finance will have agents. And managing them will look surprisingly like managing people:

The fear is “agents will replace jobs.”

Wayne’s view is closer to mine: they’ll unlock capacity for the work we never get to. There’s no shortage of problems. There’s a shortage of time.

Wayne called out a common strategy:

“Let others make the mistakes, then we’ll adopt when it’s safe.”

The problem is that it ignores the hardest part, the human adoption journey. Even if the tools mature, you can’t flip a switch and suddenly have a workforce that knows how to think, work, and collaborate with AI. That takes time.

So the smartest move is:

If you want your business to be ready for agentic AI, don’t start with the technology. Start with permission.

Permission for people to explore. Permission for leaders to admit they’re learning. Permission to make a few mistakes without turning it into a crisis. Permission to treat AI as a conversation, not a rollout.

That’s how you build momentum that survives the speed of change.

— Aamir

🎧 Listen to the Podcast Episode 1 on: Spotify | Apple Podcasts | YouTube

📘 Book: The CEO Who Mocked AI (Until It Made Him Millions) by Aamir Qutub

🚀 Dumb Monkey AI Academy

📱 Dumb Monkey AI Academy App: Apple | Android