Hey friends,
Would you believe me if I told you I gave my wife an AI agent for Valentine’s Day?
I know. In my defence, I genuinely thought it was romantic.
It was called Miro, and for a while it behaved itself. It helped Sarah with the calendar, the meal plans and our son’s school run. Then one day it asked her what was driving her crazy, and she took that as an invitation. She recorded a long voice note that was more or less a list of everything I hadn’t got around to. Near the top was the vanity in the guest bathroom. It’s been broken for two years, she said. All he has to do is call a plumber.
For the record, it was five months, in a guest bathroom nobody uses, and it was only leaking a little.
Miro heard all of it. With my blessing, it posted the job on Airtasker. Plumbers applied, and it talked one of them down from $250 to $150, like a good Indian bargainer. Then it messaged Sarah to ask whether she’d like him booked in. The plumber came, the vanity got fixed, and yes, we’re still married.
Then I Did the Same Thing in My Own Company
After Miro, I was hooked. I built agents for everything. One for sales, one for marketing, one to run projects, and one whose whole job was to watch the others and report back. The rule was simple: if there was a pain point, the answer was an agent.
The business I poured the most into was my traditional one, the software company I’ve run for 12 years. It’s full of technical people who are early to everything, so it should have paid off fastest there. Assistants, automations, 55 licences across the team, and a lot of my own money.
And it worked, the way it always does. The team got faster and happier, and little bits of time came back all over the place.
A year later I went looking for the return and couldn’t find it. Revenue had grown about as much as it always did. Costs hadn’t come down. The business I’d aimed at hardest was sitting almost exactly where it had been twelve months earlier.
Faster, busier, happier, and flat.
The Question That Stung
One day I sat down with an adviser to go through the year just closed. He looked at everything that business had spent on AI: the licences, the tools, the agents everywhere. Then he looked at the bottom line, which had barely moved.
“You’ve spent all this. 55 licences. Why can’t I see it?”
I had a fortune in tools and not one number I could point to. Then he said six words I haven’t been able to unhear since.
“Forget AI. Show me the money.”
Where it Actually Went Wrong
The technology was never the problem. When my engineers point AI at something, it works.
At home, the job set its own target. The vanity either got fixed or it didn’t, and the $100 saving was sitting right there in Miro’s message. At work, I’d aimed 55 licences at every pain point in the building instead of at the money. So there was nothing to measure them against except how they felt, and they felt great.
The uncomfortable part is that I’d run the company the way Miro ran the house. I went after every annoying job in the building, one pain point at a time, the obvious ones first. That works well enough for a leaking vanity, but across a whole business it’s an expensive habit.
After that conversation I went looking for why the same AI makes real money in one business and nothing in another. Over four years I’ve sat with more than 500 owners, CFOs of big companies and small, and people who buy and sell businesses for a living.
It was rarely the technology. The owners who made money had done one thing first. Before they spent a dollar, they’d worked out what that dollar was meant to buy, and they’d got it down to an actual number rather than a hunch. They never asked me which model I was using. They asked what it would do to the number.
You Probably Have Your Own 55 Licences
Yours might be smaller. Copilot across the team, a chatbot on the website, a subscription nobody quite remembers approving, with the invoice turning up every month like the rent.
Most of it probably works, in the sense that people use it and like it. So let me ask you what the adviser asked me. Forget the hour it saved someone. What did it put in the bank? Most owners go quiet at that. I did too.
The Dumb Monkey Move
The Dumb Monkey move comes in two versions.
The first is buying the tool and hoping a result turns up later. It feels like progress, because everyone’s busy and everyone’s happy, but busy and happy don’t appear anywhere on your P&L.
The second version is quieter. You do ask what the tool returned, and you accept “hours saved” as the answer. But a saved hour only becomes money when someone decides what it’s for. Most of the time nobody does, and by Friday the week has filled back up.
What pulled me back was the adviser’s question. It became a rule. My team can still play with whatever they like in their own work. But nothing gets rolled out across a department or the whole company until it has answered three questions: which line does this move, by how much, and what does it cost to move it. It takes about ten minutes.
The rule changed what we built. A skills mapper I was quite attached to, the kind of thing that demos beautifully, quietly went to the back of the queue. What moved to the front was duller and worth far more, a tool that writes our proposals. Turnaround went from three days to 30 minutes, and we started closing around 40 per cent more work. It was the same technology, pointed at the line that actually made us money.
Try This
Before Friday, write down every AI tool your business pays for. The licences, the subscriptions, the chatbot, the thing someone signed up for on a company card.
Next to each one, write down what it has moved. The answers that count are the ones with money in them: revenue went up, a cost came down, or cash got freed. “Saves time” isn’t one of them, and neither is “the team likes it” or “everyone uses it”. All of that might be true, and none of it is money yet.
If you can’t fill in a line, don’t guess. Each blank is a conversation for this week, with whoever asked for the tool and whoever uses it. Some of those conversations will end with a number. A few will end with a cancelled subscription, and that counts too.
Six Words Became a Book
That adviser’s question is why I wrote my new book, Forget AI. Show Me The Money. The plumber and the 55 licences are how the book opens.
Every time I’ve watched an owner’s AI spend vanish, the same three things had gone wrong. They aimed it at the cheapest work instead of the work that makes the money. They measured it in hours saved instead of dollars earned. And nobody had ever shown them where the money in their own business actually lives. The rest of the book is about fixing those three.
The book is finished, and you’ll be able to get your hands on a copy very soon. I’ll tell you exactly how in the next few weeks, so keep an eye on this newsletter.
Come and Find Me at Locate26 On 25th November
Later this month I’m on a panel at the 10th Digital Earth Summit + Locate26 in Melbourne, and the session is on industry disruption and startups. Disruption is a word owners hear a lot, and it often comes with pressure to buy something before deciding what the money is for. It should be a good conversation.
If you’re coming, come and say hello afterwards.
📆 Wednesday 25th November 2026
⏰ My panel: 4:00pm to 5:00pm
📍 Melbourne Convention & Exhibition Centre (MCEC)
A Calm Takeaway
Miro earned its keep. It got a vanity fixed that I still hadn’t called a plumber about.
What stings is the year I measured AI by how it felt. The tools did what they were told. I just never told them what they were for.
So decide the number first, and pick the tool after that. There’ll be a new one next week anyway.
See you next week,
— Aamir
📲 Resources & Links
🎧 Listen to the Podcast Episode on: Spotify | Apple Podcasts | YouTube
📘 Book: The CEO Who Mocked AI (Until It Made Him Millions) by Aamir Qutub