Everyone is adding AI to their business. Almost nobody is getting real leverage from it.

The test

• The ten-second test that tells you whether you own an AI-run business or just pay for an AI-powered one. Most owners get it wrong, and the wrong answer feels like progress while it costs a year.
• Why products can be AI-powered but businesses have to be AI-run if they really want to compete. Most people selling you AI solutions are counting on you never noticing the difference, but that difference is exactly what will end up separating the winners and losers in the AI economy.
• Why the business that put its systems in a year ago is more than a year ahead of you, and why a bigger cheque later does not close the gap. A system gets better the longer it runs on your business. Theirs has been learning the whole time. Yours would be starting from the manual.
• “Can’t I just do this myself with ChatGPT?” Yes, you can. Of course. You can also represent yourself in court, and do all the electrical wiring in your own house. Best of luck with that.

The five fronts

• The five fronts where AI-run competitors quietly pull ahead: reach, capture, convert, deliver, amplify. What each one looks like when it is working, and the tell when it is not.
• How to place your own business honestly on all five fronts in under an hour. No consultant, no vendor demo, no wishful thinking.
• Why every business review you have ever paid for measured the wrong thing, and the one measurement that shows where you are actually losing ground. A report tells you what you are worst at, not where AI is doing the most for your competitors, and they are rarely the same place... so you fix one and keep losing on the other.
• Why your worst problem and your fastest-widening problem are almost never the same problem, and which one the businesses pulling ahead fix first.
• Why the problem you notice is almost never the one costing you the most. The admin pile shouts at you every day. The enquiry nobody followed up does not make a sound, and it was worth more than the whole pile.
• What actually happens to your staff when the operations start running themselves, and why the people predicting mass redundancies have never run an expert-led business, while the people promising nothing will change don’t understand the advantages of an AI-run business.

The one system

• The one system to build first. It is almost never the one you think, and it is usually not the one that hurts most today.
• The five questions to answer before you build anything, and the one that decides whether you should be the one building it. Most owners skip that one and find out the answer on a Sunday night, three tools deep, with nothing working yet.
• Why your first system is always the hardest and your second is always faster, and what that means for which one to build first. The first one is where you find out how your business actually runs, which nobody has ever written down. The second one gets to skip that part.
• What a business run this way looks like on an ordinary Tuesday, including the exact morning digest my lead system writes, and which lines in it still need the owner.
• The one job I refuse to hand to my own agents, and why it is the point of the whole system rather than the exception to it. A true AI-run expert business understands both the strengths and the limitations of AI, and where humans still have the edge.
• Why a system that runs compounds and a tool you drive does not, and what that difference does to the distance between you and an AI-run competitor over twelve months.
• Why “my business is too specialised for this” is true, and why that is the reason it works. Nobody is touching the part only you can do. It is the booking, the reminders, the chasing and the write-ups wrapped around it that have been eating your week, and none of that is specialised at all.
• The parts of your business AI should never touch, starting with the one your clients actually pay you for.

The build, and what to refuse

• Why most of a good AI system should not be AI at all, and the one question you really must ask anyone quoting you for an AI build. Failing to ask it is like failing to ask a new hire whether they have the skills or experience to do the job... you are setting yourself up for disaster.
• What a system should do with the enquiry that does not fit the form, and why the one that guesses will cost you a client before it saves you an hour. Every business gets the odd one. A system built for the normal case handles it confidently and wrongly, and the first you hear of it is the client.
• The one page any system running your business should be able to show you on a Tuesday afternoon, in language your bookkeeper could read. Ask for it before you sign anything. If the builder cannot produce it, you are about to pay for something you will never be able to tell is working.
• Whose name should be on every account your system runs on, and the one question that tells you whether you bought a system or rented a dependency. If switching it off means ringing somebody else, it is not yours, whatever the invoice said. Ask anyone who has tried to leave a platform with their client list still inside it.
• Why the last thing you tried did not stick, and why it was not you: you were sold a tool you had to operate and told it was a system that would run. That is a treadmill sold as a personal trainer. Of course it is in the garage.
• How to tell agency theatre from a real build in the first meeting, before any money moves. You will learn more from the hesitation than from the answer.
• What it costs to build one of these now, against what it cost when it took an engineering team, and why most owners are still carrying the old number in their heads. That old number is what stops you starting, and it suits everyone still charging it.
• Why a business with real revenue and real deadlines is the right place to test this, and what broke when I did. Everything in this book ran inside my own businesses before it went anywhere near yours, and not all of it worked first time. You get the parts that broke (and were fixed or removed) as well as the parts that worked, because that is the half most people selling you AI will never show you.
• What to do first if you are not technical: what to build, what to rent, and the one thing you should refuse to buy.