← Half a Person

Struggling at the top

The shortcut that left me further behind

10 September 2026

I had one of those rare days at work: nothing in the calendar, eight hours that were actually mine, and a clear runway to build the marketing roadmap my team has been asking for - the one that I've been meaning to sit down and write for weeks. I started well, or what felt like it. I decided that the smart move was to get the system right first - the template, the structure - before I put a single real idea into it. And of course, AI seemed like the obvious way to speed that part up.

So, head down, headphones on, I got stuck in. I was flowing. Wow this felt this good.

And by the time I looked up... well, most of the morning had gone. Not to the roadmap itself but to the search for the right way to build one. I tried one approach, didn't trust it, then tried another. Somewhere in there I noticed the advice coming back wasn't actually tailored to what I was trying to achieve. It was generic enough to fit any team's roadmap, which meant it would fit no one's actual roadmap, and certainly not mine. I switched AI tools. That didn't fix it either. It gave a different opinion on my approach to the structure, and bad-mouthed the suggestions of the first LLM, but still no credible solution.

The rest of the day went as most days do: people needed things. Someone on my team needed help with something urgent, and later, unrelated, someone else did too. Then I had to squeeze in the school run. Then another urgent call that ate up the last of my hours. That's not unusual and it isn't really the story here. What I had left to show for the day once it ended was three lines. Three lines. Three lines, on a roadmap the business keeps asking for, after a day with no meetings and every intention of finally getting it done.

The easy explanation is that I'm bad at this now, that I've somehow lost the ability to do work I used to be good at, simply because I let a tool anywhere near it. That's not quite it. I've built roadmaps before without any help from AI at all, decent ones, and it never took me a morning just to settle on a shape. Reaching for the tool this time cost me more time than doing nothing would have. If I'd opened a blank document and just started, badly, I'd have more than three lines by now.

What was actually happening underneath all of it is a habit I recognise, and it's older than any AI tool. Given real work that needs my own understanding, something I already had the material for in my notes and my head, I went looking for something outside myself to do it instead. It used to be a colleague's opinion, or a framework borrowed wholesale from somewhere else. This time it was an LLM. It's the same instinct that's kept me employable for twenty years in a job built on being articulate and quick with other people's ideas: find the version of this that's already been worked out somewhere, rather than sit with the slower job of working it out myself. It's a good trick, and it's carried me a long way. It just isn't thinking, and the roadmap needed thinking. I need thinking, in order to understand.

Building the system was its own kind of trap, because it felt like doing the work - professional procrastination you might call it. Comparing two AI outputs, weighing one template against another, is effortful, and it leaves you with something to point at: a structure, an approach, evidence you did something today. None of that requires actually knowing what belongs on the roadmap. That comes from reading my own notes and plans closely enough to make real decisions, and it doesn't give you anything to compare or choose between while you're doing it, which is exactly why it's the part I skipped. Infrastructure is satisfying to build. It can also be a way of not yet starting.

I don't think the fix is simply avoiding AI on the next one of these. I think it's catching the moment, right at the start, where I reach for something outside myself before I've spent even ten minutes with what I already know. Today that cost me most of a day and left me further behind than if I'd never opened the thing at all, and I still don't know how to catch it before it happens again. I just know it isn't new, just like I know it's not really AI's fault.