Hey friends,

An owner I know sat me down with a straight question. He'd bought one AI for his sales team, another for accounts, a note-taker for his meetings and a tool that read his invoices. Every one of them worked. Someone had just shown him something clever for scheduling, and he wanted to know whether to buy that too.

What he was really asking was whether more tools would ever add up to a business that ran better. I had to tell him no, and the clearest way I know to explain why is our own tenders.

A single tender used to take us six to seven full days. Now it takes one. People always ask which tool did that, and the tool is the least interesting part of the story. An agent watches the tender portal and scores each opportunity, the documents get turned into a clear brief, and a first draft gets built from submissions we've already won before my team shapes the final answer. Every step draws on what the business already knows, and a person still owns the result.

His business is probably a lot like yours. You buy AI the way you buy tools, one at a time, for one job at a time. A year later you have ten clever tools and a business that runs no better, because none of them talks to another, none of them knows what the business knows, and stitching them together still lands on you. What you actually bought was ten more things to check. The thing worth building is the system that ties them together and stays put while the tools underneath keep changing.

Think about the phone in your pocket. It used to be five separate objects: a phone, a camera, a calculator, a diary and a map. What made it impossible to put down was never a better camera. It was the operating system underneath, turning all of it into one thing you just use. A business is never quite that tidy, but the shape holds.

An AI operating system has five layers, each one standing on the one below.

At the base are your people, and everything above them exists to take load off them. Most of the industry will tell you AI is here to replace your people, and I think that gets it backwards. It's there to lift the parts of the day that were never the reason you hired them, and to catch what a tired person misses at four in the afternoon.

Above them sits the institutional memory, where what your best people know gets out of their heads and into something the business keeps. It's also the layer a buyer pays more for, which is why I'd keep it in something you own and well away from the inside of any one AI model.

On top of the memory sits an assistant, one AI in everyone's hands that knows your products, your prices, your policies and the way you do things. An assistant is only as good as what it can draw on, though. Point it at nothing and you've got a stranger with good manners. Point it at the memory and it starts answering the way your best manager would.

Once the business knows things, it can start doing things, and that's the workforce. You'll hear it called agents, but what matters is what the job needs. Some jobs need a person with AI alongside them, some need the whole thing done and handed back, and some need a worker that runs on its own and only comes to you when something's off.

One client had an audit bearing down and a senior person losing weeks to incident report forms. So we turned the reporting into a phone call. After an incident you dial a number, and a worker on the other end walks even a rattled person through the questions, files the report, escalates it and keeps the audit trail. A human still checks the end of it, and the senior person got their weeks back.

At the top of the stack sits the control room. It watches everything below it, and it's the layer a pile of tools will never give you.

You can hand off every task in the building and still be stuck at your desk, because the last job keeping most owners there is the watching rather than the doing.

A facilities group my team worked with had its information spread across five places, from the finance system to whatever lived in people's heads. The only way the owner could know whether a contract was in trouble was to go and ask. So he asked all day, and still got blindsided.

We connected the lot and put a traffic light on every contract across quality, cost, delivery, management and safety. When a contract started drifting towards red, the right person got a message on their phone straight away rather than in a report at the end of the month.

Then it started noticing things nobody had built a chart for. An area manager who'd quietly stopped turning up to the weekly meetings. A run of complaints that all traced back to one subcontractor. You only ever chart a problem you already know you have. What you want is the business telling you when it needs you and staying quiet the rest of the time.

There are two ways to get this wrong. The first is the one the owner with the scheduling tool was about to fall into: buying the next clever app, and the one after that, and calling the pile a strategy.

The second is the opposite, and it costs more. You hear "operating system", decide you need all five layers at once, and pay for a platform you were never going to use. A friend of mine runs a small, newer business and was losing his week to an advisory board asking for information he had to pull together by hand. I could have sold him the full connected platform. I told him to connect his AI to the few good tools he already used, build one simple skill, and ask it live whenever the board had a question.

Build what fits the business in front of you, and be suspicious of anyone whose answer happens to match the invoice they'd love to send you.

Two questions. First, what does your business know that lives only in people's heads, and would hurt most to lose? Write it down, with the name of the person who'd take it with them if they left.

Second, what's the one thing you check obsessively because you don't trust it to tell you when it goes wrong? The cash position, a big contract, the supplier who always slips. That's the first thing worth wiring up so it taps you on the shoulder instead.

Next month I'm taking all of this into a sector where the people layer is under more strain than almost anywhere. Employment and community services run on people and paperwork, and they hold information about people at some of the hardest moments of their lives.

On Wednesday 11th November I'm speaking at IEP Connect Melbourne, a half-day workshop on practical AI for employment and community services, hosted by the Institute of Employability Professionals Australia. It's a hands-on afternoon on how AI can take admin off the people doing this work while keeping them at the centre of it, with short showcases, live demos and real conversations about the trials and pilots people are running. Bring the process your team complains about most, and come and find me afterwards.

📆 Wednesday 11th November 2026

⏰ 12:00pm to 5:00pm

📍 Community Hub, 912 Collins Street, Docklands

🎟️ Tickets: https://www.tickettailor.com/events/theinstituteofemployabilityprofessionalsaustralia/2215269

I keep coming back to that owner and his scheduling tool. He came to me asking which tool to buy next. The question underneath it was who holds all of this together, and for most owners the honest answer is still them.

Every layer in this newsletter takes something off you. The memory keeps what your people know, the assistant puts it in everyone's hands, and the workforce does the jobs that used to eat your best hours. The control room takes the last thing, the worry. That's what keeps most owners chained to the desk long after the tasks have gone, lying awake as the only person who'd notice if it all went wrong.

None of this gets installed on a Tuesday afternoon. Start with your people and what's stuck in their heads, and add the next layer once the last one is earning its keep. The tools underneath will keep changing every few months. Build the thing that holds them.

See you next week,

— Aamir

🎧 Listen to the Podcast Episode 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