The Porsche, the Spreadsheet, and What AI Taught Me About How I Make Decisions

I should tell you upfront that this post is ostensibly about a car. But it's not really about a car.

It's about the way I make decisions. And how I used AI to catch myself doing it badly.

The research that went sideways

I've had a long-standing interest in Porsche 997s, specifically the 997.2, the second generation of the air-cooled-adjacent era that sits in a sweet spot of performance, usability, and ownership character. For a while now, it's been one of those "one day" items. Present on the mental list, not actively pursued.

During my AI audit phase, I decided to use the 997 research as a test case for AI-assisted decision-making. Not because I was ready to buy, I wasn't, but because I wanted to see how I behaved when I ran a real desire through a structured AI process instead of my usual approach, which is: read forums, form an opinion, confirm it enthusiastically, and call that research.

I gave the AI my situation: income profile, financial goals, the 2027 home purchase target, current debt reduction focus, and then asked it to help me think through the decision honestly.

What it found in my reasoning

The AI didn't tell me whether to buy the car. It did something more useful: it mapped how I was thinking about the decision.

It pointed out that I was treating the car as an isolated financial decision when I'd described it as emotionally significant. That I was benchmarking the cost against a monthly increment rather than against the 2027 goal I claimed was my priority. That I'd front-loaded my research with enthusiasm (forums, spec comparisons, model year differences) before I'd done any structural thinking about whether this was the right time.

It wasn't harsh. It wasn't telling me not to buy the car. It was showing me the shape of my reasoning, which is the thing I most consistently fail to examine when I want something.

The pattern beneath the purchase

What I realised, and what I've been sitting with since, is that I have a consistent decision-making pattern that I'd never labelled before. I do thorough, intelligent-feeling research on the emotional dimensions of a decision (what's the right spec, what do experts prefer, what's the community consensus) and I do very little structural thinking about the decision itself (is this the right time, what am I trading off, what does this tell me about my priorities).

I optimise the object of desire while underanalysing the desire itself.

That's not a car problem. That's a me problem. And I needed AI to show it to me because I'd been successfully hiding it from myself behind good research habits.

The magic of my modelling guided me to design a transaction that kept the passion alive while maintaining a sense of practicality. I viewed dozens of cars, much to my mate's frustration. But what emerged was a transaction-and-decision design that ensured I'd pick a car that was both a passion and a purpose. It helped me set a decision criterion for a purchase that would actually appreciate.

Practical AI: Pressure-test a real decision

Think of a decision you're currently sitting on. Something real, not hypothetical. A purchase, a career move, a conversation you're avoiding, a commitment you're considering.

Write a paragraph describing where you are with that decision right now, what you're thinking, what's holding you back, and what's pulling you forward. Paste it into your AI and use this prompt:

"Based on how I've described this decision, what does my reasoning pattern suggest about how I'm approaching it? What assumptions am I making that I haven't examined? What's the question I seem to be avoiding?"

The goal isn't to let AI make the decision. The goal is to see your own thinking more clearly than you can from inside it. Ask it what additional sources of information might help to make a better decision. Ask it to ask you follow-up questions to gain further clarification.

The outcome might just surprise you.