He was about to wire the money. The opportunity looked clean, the space was familiar, and the excitement was real. Then the AI he had been using to stress-test his deals delivered a cold, ego-free assessment: the largest AI companies in the world were almost certainly going to enter that personal finance tool market, and competing with them would be a losing fight. The investment never happened. ‘I literally could have lost millions of dollars making that decision,’ he said later. That near-miss is not a cautionary tale about AI. It is, for the self-described number-one angel investor in Canada who has backed founders alongside those behind Google and Jay-Z, the whole point.
Why the first move is letting the money leave the building
The framework starts with a philosophy he has held for years and now executes faster than ever with AI at his side. Wealth preservation, in his view, is a trap. ‘Preserving wealth is like building a wall around your money,’ he explained. ‘It blocks other money from coming into your life.’ Once that mental shift is made and a person decides to deploy their dollars rather than defend them, the harder question arrives: where, exactly, do those dollars go?
His answer involves two specific prompts run inside Claude, Gemini, or ChatGPT. The first asks the AI to map out the second and third-degree winners and losers inside any industry or trend, surface the suppliers, the adjacent industries, the companies about to be disrupted, and then return ten names a human investor would not have thought of on their own. The second prompt asks for an analysis of recent market signals around those names and pushes for another layer of second and third-degree plays. His example is direct: ask AI about the AI industry and it will not tell you to buy Nvidia. It will start talking about energy infrastructure and the electricians needed to install it.
There is a hard constraint built into this process. He only runs the prompts inside categories where he already has deep, personal context. He made a painful $100,000 mistake buying homes in Detroit years ago, properties that rented for as little as eight thousand dollars each, properties that turned out to be unlivable and never properly managed. The reason it happened: he invested in something he did not know. ‘Sometimes pain is the perfect teacher,’ he said. The rule that came out of it is now one of three mandatory filters: do you have an unfair information advantage, will the investment command enough of your attention to actually manage it, and will the thesis still be true in ten years?
Red teaming the deal before the money moves
Once a promising opportunity clears the filter, it gets pressure tested rather than celebrated. He prompts the AI to play out the worst-case scenario and produce ten reasons the investment could go to zero, with explicit instructions to skip the diplomacy. Each objection becomes an item on a checklist. If he cannot answer most of them with high confidence, he passes on the deal. He runs the same stress-test prompt across two or three different AI models, because different systems surface different blind spots.
The monitoring phase is where the system becomes genuinely self-sustaining. Every investment he holds, real estate, stocks, company valuations, vehicles, lives inside a single AI-built dashboard. It tracks price movements, flags red, yellow, or green health signals for each asset, and sends alerts when something moves by more than ten percent. He reviews the whole board every Monday morning. ‘Paying attention to where your money goes is how it grows,’ he said, citing Peter Drucker’s line that what gets measured gets managed as the principle behind the daily habit.
His most recent deployment of the framework led him to Hark Labs, a company building what he described as an embodied AI model designed to power the robots of the future, with the founder of one of the leading robotics companies as its lead investor. He applied all three filters: he knows the AI space, he wrote a large enough check to stay focused, and he believes the direction the world is heading makes the thesis durable over a ten-year horizon.
The dashboard that runs on a Monday morning
The specific prompt he uses for the monitoring layer asks the AI to build a live dashboard that tracks every asset on the list, displays how healthy each one is, watches for price movements against market valuation, and flags anything that needs a closer look.
Once, he asked a friend how much money he had sitting in his business. The friend said, ‘About 70 grand.’ The answer stopped him cold. ‘I can tell you to the penny how much I have today,’ he said, ‘because I monitor it.’
A green dashboard on Monday means the week starts clean. A red one means digging in.
