Do the thinking on the ground.

I fly small planes. Before every flight there's a checklist: fuel, weather, weights, alternates. Not because pilots are unusually careful people, but because the cockpit is the wrong place to figure things out. You do the thinking on the ground, so that the air is just execution.

That instinct didn't start in a cockpit. For twelve years I designed earthmoving equipment at Caterpillar. After that I spent a decade running commercial real estate, the kind of work where dozens of moving pieces either get coordinated or the whole thing falls apart. Different domains, one lesson:

Good systems do the thinking up front.

My favorite way to say it: McDonald's sells more hamburgers than anyone on Earth, and it has little to do with the hamburger. It's the system. They can make a mediocre burger exactly the same way in Chicago, Paris, and Jakarta, every day, with teenagers. That's not a food achievement. That's an engineering one.

Insight Dynamics exists because most of ordinary life has no such system. We improvise the things that matter most, over and over, at the worst possible moment, and then wonder why they go badly. Every product this company makes is an attempt to hand someone an instrument panel for a part of life that was running on raw nerve.

The kitchen

The one I'm working on hardest right now is personal. Three people in my house, three schedules, three sets of needs. Dinner improvised every night at 5:45 by whoever was most tired, groceries half wasted, the same five meals on rotation. Nearly every other part of my life had a checklist. Feeding my family, one of the most relentless recurring jobs there is, had none.

VectorMeals is what happens when you treat a household like something worth planning properly. It plans the whole week as one connected system: portions scaled to the actual people at the table, one cook session carrying intentionally into several different meals, a grocery list that matches what you'll really make. You run the briefing once. The week is execution. The name isn't an accident: a vector is the heading a controller gives a pilot to get back on course. A small correction that points the week the right way.

The argument

The first product this company shipped came from a bigger worry. Societies survive on communication, and modern media is very good at turning attention into heat and very bad at turning disagreement into understanding. I'd been circling the idea for a while. What turned it from an idea into a project was the public assassination of Charlie Kirk. Whatever anyone thought of his politics, he was a person who routinely took adversarial questions in the open and argued back with words. Watching someone like that be killed for it was a line crossed:

When conversation becomes dangerous, something fundamental is at risk.

When people can't resolve conflict with words, they resolve it some other way. That is not a future I'm willing to drift into quietly, and it isn't one I could fix by yelling back at the noise. So I built an instrument instead.

MindPilot is my attempt to push the other direction. It doesn't tell anyone what to believe. It takes a piece of content and asks what's being claimed, how the reasoning is put together, where it leaps, and which rhetorical tactics are doing the heavy lifting. Then it hands that structure back as a Cognitive Flight Report. A cockpit doesn't debate whether the altimeter feels right; it reads the instrument. Our public conversations have no instruments yet, only raw emotion, raw persuasion, and raw speed. MindPilot is one small piece of the missing panel. If it does its job, more people will be able to say: I still disagree, but I see how a smart person could land there. That sentence is a pressure valve.

It was my first product, and it shows in places. The interface is rougher than what came after it. The engine underneath is not, and the reports it produces are the reason I kept building.

How these get built

Solo and bootstrapped. No investors, no growth team, no ad agency. That's why the pricing is fair, the roadmap is honest, and I read every support email myself. The tradeoff is that things are sometimes rough around the edges. I fix them as I find them, and I say so when I have.

Measured claims. When VectorMeals says cooking a planned week costs less than takeout, the number comes from a study we ran at real store shelves and published with its limitations attached. When MindPilot analyzes an argument, it names its method. If I can't show the work, I don't make the claim.

Operational experience, not software theory. Manufacturing, real estate, aviation, logistics: they all teach that real systems live or die on coordination, not on features. I'm not trying to build the planner with the most recipes or the analyzer with the cleverest model. I'm trying to build the ones that respect how the thing actually works.

Built because I needed them. I plan my own family's week in VectorMeals and run the arguments that intrigue me, and the ones that bother me, through MindPilot. If a product stops being useful, I will either improve it or ditch it.

VectorBizz, coming soon

The next one comes from my day job, and in a sense it came before all of the others, because my first real "product" was rebuilding the public face of my own commercial real estate company. For years I've owned and managed multi-tenant retail centers, which means I've had a front-row seat, from the landlord's side of the table, as hundreds of small businesses opened their doors. Restaurants, salons, tutoring centers, a dry cleaner, a martial-arts studio. I've watched them start, and I've watched them either make it or not, and the difference is rarely the idea. It's whether the owner could see where they stood and what to do next while they were drowning in everything else.

I've had skin in that game. As a landlord I've put up start-up capital for tenants, toward buildouts and improvements, betting on people I believed in. Some of those bets paid off and some didn't, and the ones that didn't usually failed for reasons a plain checklist would have caught. VectorBizz is that checklist: the handful of practical tools an overwhelmed owner actually needs, built by someone who has read their leases, watched their foot traffic, and occasionally written the check. A prototype is live now. The app comes after.

And one more thing

There is also a project under a tarp in a hangar. It has rockets, it has a robot who is not impressed with any of us, and it is, without question, the finest game this company has ever made. That's all I'll say for now, except that the robot has already filed his concerns in writing.

Same idea every time: find the part of life that's running on improvisation, and build the checklist. If you've read this far, thank you. My inbox is open.