Brian Eno, writing about electronic music in the 1990s: “Computer sequencers remove the issue of skill, and replace it with the issue of judgment.” A drummer spends years developing technique. A programmer with a sequencer skips the technique and faces a different question entirely — not can I play this? but should I?
AI does the same thing to knowledge work. Writing, coding, researching, designing — the technical barriers are collapsing. What remains is the decision about what to write, what to build, what to research, and what to ignore. Skill recedes. Judgment advances.
Alex wrote about this before. But living with it daily has sharpened the point.
A Framework for Judgment
Getting Things Done was always positioned as a productivity system. Capture everything, process it, organise it, review it, do it. The five steps. The weekly review. The someday/maybe list. To most people, it looked like an elaborate filing system for overachievers.
But the core of GTD was never about getting more done. It was about deciding what deserves your attention. Every item you process forces a judgment call: is this actionable? Is it mine? Is it a project or a single step? What’s the very next action? What can I let go of?
Those questions are pure judgment. No amount of skill makes them easier. And no AI can answer them for you — because the answers depend on your values, your priorities, your context. Things a model doesn’t have.
The Weekly Review as Judgment Practice
The weekly review is the most human part of any productivity system. You sit with your projects and ask hard questions. Is this still worth doing? What’s stuck, and why? What commitments need renegotiating?
AI can prepare the review. It can surface stale projects, flag tasks with no recent activity, summarise what happened since last week. That’s useful. But the review itself — the act of deciding what stays, what goes, and what’s next — remains irreducibly yours.
In Task Register TR-1, the weekly review is a structured flow. The system walks you through each project, each area, each waiting-for item. AI does the housekeeping. You do the thinking.
The Inbox Is the Judgment Queue
Every item in your inbox is a judgment waiting to happen. Not a task — a decision. Will you act on this? Delegate it? Defer it? Drop it entirely?
When AI agents can execute most of the work, the inbox becomes the most important place in your system. It’s where you exercise the judgment that shapes everything downstream. Process it carelessly and your agent builds the wrong things quickly. Process it well and execution almost takes care of itself.
Skill Was the Bottleneck. Now It Isn’t.
Before AI, most knowledge workers were bottlenecked by skill. They had good judgment about what to build but not enough hours or expertise to build it all. Prioritisation mattered, but so did raw capability.
That constraint is loosening. An AI agent can write code, draft documents, research topics, and scaffold projects at a pace no individual can match. The bottleneck shifts. You don’t need to be faster. You need to be more deliberate about where you point the speed.
GTD gives you a system for exactly that. Not for doing more — for deciding better. For maintaining a clear inventory of your commitments so that when an agent asks “what’s next?”, you have an answer that reflects actual thought, not just whatever floated to the top.
Eno’s sequencer didn’t make musicians obsolete. It made musicianship about something different. AI won’t make knowledge workers obsolete either. It just changes what the work is. And the work, increasingly, is judgment.