Greetings Reader

So you got the email a few weeks ago. Maybe a few months ago. Something along the lines of "Exciting news!" with a screenshot of a chatbot interface and instructions to log in to Copilot, ChatGPT, or whatever your organisation's AI of choice might be.

And then… nothing. No training. No guidance. Just you, a search bar that says "Ask Anything!!!" and an expectation of dramatically increased productivity.

If that's you, you're not alone. So many companies hand out paid AI licences and then just… leave people to it. Like it's Google. Like you'll just figure it out by typing "make my email sound more friendly" or "summarise document".

I keep hearing from leaders: "Yeah, people are using it for their emails and stuff." And every time I hear that, a small part of me dies. That's not a productivity win. That's an AI slop win.

Because this is what typically happens. Someone takes three dot points, dumps them into Copilot, and gets back a 400-word email that is clearly written by AI (littered with em dashes and suspiciously correct grammar). The recipient then wades through all that slop to extract… the original three dot points. Except by now, they're a different set of three dot points. Nobody wins. Everybody's annoyed.

So: what should leaders actually do? (And if you are not a leader, forward this email onto one...)

1. "AI will make us more efficient" is the wrong pitch

The organisations getting this right aren't leading with "AI will make us more efficient". Because what people hear is "AI will replace me and I should probably update my LinkedIn."

Instead, treat people like humans: We know you're overloaded. And AI can help.

Most knowledge workers are already stretched (the ones we work with sure are). Doing 50+ hours when they're paid for 40. Buried in admin they didn't sign up for.

So frame it that way (assuming you mean it): We're not bringing in AI to get rid of people. We're bringing it in so you can stop drowning in administrative rubbish and actually do the work that matters. Maybe even leave on time for once.

2. Stop assuming people will figure it out

I continue to be surprised by how many organisations hand out paid AI licences and then sit back waiting for the productivity gains to come raining down like confetti.

They assume it's intuitive. It's not.

Yes, there's a friendly chat interface with a prompt box and a go button. Very inviting. Very simple. Tells you absolutely nothing about how to actually use it well.

So people learn two or three tricks (rewrite this email, summarise this document) and that's where they stay. They've bought a Thermomix and they're just making smoothies.

The stuff that genuinely shifts productivity? Using AI to challenge and pressure test your thinking. To sharpen your work. To automate repetitive workflows. To do the shallow work autonomously so you can do the deep work.

But people need to be shown how. They won't stumble into it on their own (or at least, most won't).

3. The missing piece that turns AI training into AI adoption

This is the one that separates the organisations that get genuine ROI from AI and the ones that spent a lot of money on licences nobody uses properly.

Identify your change leads. Maybe that's team leads. Maybe it's dedicated people. Either way, give them more training than everyone else - not just "which button to press" training, but "how does this tool actually fit into someone's real workflow" training.

Most staff are not tapped into AI best practices. They don't know all the capabilities. They just know they have a new tool and approximately 47 other things competing for their attention.

A good change lead sits down with someone and says: "You write that weekly report every Friday? Let me show you how AI can cut that from two hours to ten minutes." That's the real unlock. Not generic training. Specific, workflow-level support.

In short...

If you're a leader rolling out AI (or about to), here's your checklist:

1. Nail the why. Make it about reducing overload, not replacing people.

2. Build capability. Train people properly. AI is not intuitive to use well.

3. Train change leads. Give specific people the skills to help others fit AI into their actual workflows.

Skip any of these and you risk the worst outcome of all: staff who actively resent the tool, an expensive set of licences gathering digital dust, and a productivity gain of precisely zero.

(Or worse still - a productivity loss. Which, as we've established, is entirely possible when untrained people start generating AI slop that everyone else has to wade through.)

If you're reading this thinking "oh no, we've done none of this", that's fine. That's many organisations right now. But the gap between companies who get AI adoption right and those who don't is only going to widen.

And if you need help, just reply to this email if you want to chat or check out Inventium.ai (we have helped heaps of organisations with AI roll-outs). My part-time job has become speaking to senior leaders who are grappling with this stuff and I love nothing more than sharing how to do it well.

Cheers

Amantha