The Stuff That Actually Runs

The prototypes get all the attention. They’re the fun bit, and they make the better story. But if you actually want to know what six months bought me, don’t look at the half-built apps. Look at the dull, unglamorous stuff I now use every single day and would genuinely struggle to work without.

It’s a system, not a drawer of tools

The thing I’m proudest of isn’t any single skill. It’s that they connect. I didn’t end up with a pile of clever one-off gadgets, each needing to be remembered and wound up by hand. I ended up with an operating system for my working life, where the parts hand off to each other.

That distinction is the whole game, and it’s the one most writing about AI tools skips. A drawer of tools is a chore you have to keep opening. A system does the carrying for you. One of those changes your week. The other just adds to it.

How I run client work

I now have a repeatable way to walk into an organisation and map how its technology and operations actually fit together. A workflow for taking an opportunity from the first introduction, through each conversation, to a clear decision. A library of the reusable parts of a consulting engagement, so I’m never starting from a blank page.

None of those is individually clever. The point is that “do a technology audit” went from a daunting, bespoke, start-from-scratch effort to something I can run to a known standard, the same way, every time. Consistency is a capability in its own right, and it’s the one clients actually pay for.

How I set direction and keep it

There’s a proper goal-setting structure sitting underneath everything: one genuinely important goal per part of the business, with the handful of weekly actions that move it and a scoreboard I can’t hide from. On top of that, a weekly review that actually happens, and a couple of skills whose entire job is to argue with me and check whether the plan survives contact with reality.

Strategy usually dies in the gap between the away-day and the following Monday. Mine doesn’t, and not because I’m disciplined. It’s because holding the plan is now a routine the system runs, rather than a thing I have to remember to care about. Take the remembering out of it and consistency stops depending on my mood.

How the work actually moves

Everything lives in one task system, and on top of it sits a small crew of agents. One picks up the next job. One turns a rough task into a proper brief. Others grind through batches of work while I’m doing something else entirely, or asleep.

Somewhere in there I stopped being the person who does the work and became the person who supervises it. That shift, from doing to directing a very cheap and very fast team, is the change I didn’t see coming and the one that matters most. It’s also the skill I think is about to be worth the most: not prompting, but managing.

How this post got made

What you’re reading went through the machine too. One skill drafts in my voice. Another strips out anything that sounds like a chatbot wrote it. A third publishes it.

I’m using the system to describe the system, which sounds like a gimmick but is really the test. If the thing couldn’t produce a clean piece of writing about itself, it wouldn’t be working. It can, so it is.

So what does “working” actually mean?

Not that I’ve automated my life. It means the unglamorous plumbing, the connective tissue nobody puts in a demo, is the real output of six months, and unlike a prototype it compounds. Every new thing I start now plugs straight into it.

And here’s the part that matters for what comes next: none of it is a toy I leave behind when the sabbatical ends. It’s portable. When I walk back into client work, I’m not bringing a nice story about AI. I’m bringing a working operating system, and the ability to build one for someone else.

The prototypes were the fun. This is the asset.

Posts in this series