Thinking out loud.

If you want to guess what the next seven years of AI will look like, you really only have to answer two questions: do agents get reliable enough to trust on their own, and does AI start to make AI better? A seven-year scenario map for 2026–2033 — four futures, the order they separate in, and the quiet signals that tell you which one is coming.

Most things get more complicated on their own. Features accrete, steps get added, processes grow a new exception every time something goes wrong. Nobody decides to make it complicated — it just happens, the way a desk gets cluttered, one reasonable addition at a time. Simplicity is the only state that takes active work to hold.

Early in my career I measured a good day by how much I personally got done. Tickets closed, code written, problems handled. It felt productive, and it was — for one person, for one day. It took me a long time to notice that the people who actually moved things weren't doing more work. They were doing work that did work for them.

Spend enough time around factories and you stop believing in general improvement. A manufacturing line has exactly one station that sets the pace for the whole thing. Speed up any other station and you've accomplished nothing — you've just piled up more work in front of the slow one. Until you fix the bottleneck, nothing else you do matters.

I deploy autonomous agents into enterprise workflows for a living, so I watch a lot of AI demos. Most of them are beautiful — a clean prompt, a clean answer, a room full of people nodding. And most of them would not survive ten minutes inside an actual company.

For most of my career, making things was the expensive part. Writing the code, producing the analysis, drafting the document — that's where the time and the money went, so that's what we optimized and what we paid for. That era is ending. I can now produce a first draft of almost anything in minutes. So can everyone else.

I caught myself mid-conversation with a client the other day realizing the point I was making wasn't even mine. I'd read it in some strategy piece, absorbed it, and repeated it like I'd come up with it. When your entire job is original thinking applied to hard problems, that should concern you.

I used to think life was about doing something big. And in a way, I still do. The urge to build something that leaves a mark hasn't gone away. But I had a picture in my head of what success looked like, and it was always on a big stage, with a sea of faces watching. I made choices based on how they would look to that audience, on what would earn the loudest applause.

Running a company is mostly a long stretch of not knowing. You don't know if the product is right, if the client renews, if the cash lasts the quarter. Early on I treated all that uncertainty as a reason to be careful — to hedge, to wait for a little more information before I committed to anything.

Integrity is what you do when no one is looking and you know no one is looking. Not when a camera might catch it. Not when it might come out later. When there's genuinely no audience and no consequence — that's when you find out what someone actually is.

Every morning I face the same small fight: do I write the code, or build the thing that means I don't have to? Close the deal, or build the system that closes deals? Working *in* the business is the actual work — shipping product, talking to customers. Working *on* it is the slower stuff — hiring, process, systems. Both are real work. Getting the mix wrong has killed more of the companies I've watched than any competitor did.

Everyone's hunting for the hack — the routine, the tool, the framework. After 20 years and a pile of shipped and half-shipped projects, I can tell you the only one that's ever worked for me: I started.

This is what I do for a living now, so I'll be blunt about it: most companies pay expert salaries for robot work. Analysts copying numbers between spreadsheets. Engineers writing status reports. Smart people spending the bulk of their week on tasks that don't need anyone smart at all.

For most of my career I ran big teams. Data orgs at banks, platform groups at large companies, onshore-and-offshore setups with more people than I could keep straight. Now I run small ones. Small wins, and it isn't close.

We talk about discipline like it's the whole game — grit, willpower, pushing through. I've come to think enthusiasm is the more honest lever. Everything is easier when you actually want to do it, and most of the productivity advice people collect is really just a workaround for doing things they don't.

I spent a lot of years in places where a "fast" project took two quarters. Now I ship things in days. The difference isn't that I got smarter. I just stopped letting work take as long as it wanted to.

"Work expands so as to fill the time available for its completion." Cyril Northcote Parkinson wrote that in 1955, and once you've seen it you can't stop seeing it. It's why the two-week task becomes a two-month project, and why the same team that shipped weekly as a startup ships quarterly once it's big.