Your To-Do List Just Became Multi-Threaded

This morning I had five browser windows open before 7 a.m. One was this post. Three were AI agents chasing down research threads I'd kicked off the night before. The fifth was reorganizing a client directory I hadn't touched in two years.

That's five to-do items running simultaneously. And I'm no longer the throughput.

For 23 years, Getting Things Done has been the operating system for knowledge work. David Allen's framework — capture, clarify, organize, reflect, engage — assumed one human, one brain, one task at a time. "What's the next action?" was the question that organized everything. Your productivity was a function of how efficiently you could sequence your attention.

That assumption just broke.

The Single-Threaded Assumption

I've been using GTD since 2003. Religiously. Context lists, weekly reviews, the whole apparatus. It worked because the constraint was obvious: me. One pair of hands. One context at a time. The methodology optimized for moving serially through a list without dropping commitments.

GTD was built for a single-threaded processor in a world that's now parallel.

When I'm working with AI agents — and if you're reading this, you probably are too, whether you call it that or not — the constraint isn't my attention anymore. It's my decision-making. The agents don't wait for me to finish one thing before starting another. They run. I pointed three research agents at different topics last night. By morning, all three had drafts waiting. None of them blocked on each other. None of them needed me until they hit a decision point or produced something worth reviewing.

The to-do list didn't disappear. It split into three categories that GTD never anticipated:

1. Threads in flight (who's working on what right now)

2. Decisions waiting for you (your actual bottleneck)

3. Outputs to review (the part GTD didn't have a slot for)

What Happened to Media Is Happening to Productivity

We've seen this pattern before. When newspapers moved online, the constraint flipped from print deadlines to attention. You could publish unlimited articles — the question became which ones anyone would read. The bottleneck moved from production to distribution.

Knowledge work is having its digital-publishing moment. The constraint is no longer how many tasks you can execute. It's how many threads you can supervise.

I'm watching this with clients right now. A CFO I'm advising has junior analysts who used to spend three days building board decks. Now they kick off an agent Friday afternoon, review a draft Monday morning, and spend their time on the analysis that actually requires judgment. The analysts aren't working less — they're orchestrating more. Their calendar looks nothing like their task list used to.

The traditional productivity question was: "What should I do next?" The new question is: "What's running, what's blocked, and what just finished that I haven't looked at yet?"

The Review Queue Nobody Planned For

Here's the part that's breaking people: the output queue.

GTD had "waiting for" lists to track things you'd delegated to humans. Those humans were slow. You checked in weekly. Agents are fast and cheap enough to spin up five of them before breakfast. The review queue is the new inbox, and it fills faster than the old one ever did.

I was talking to an auditor last week who'd automated evidence gathering across three workstreams. He was drowning. Not because the agents failed — because they succeeded. He had three polished reports sitting in his review folder and no system for prioritizing which to read first, which to ship, which to iterate on. His GTD context lists had no category for "things that are done but I haven't verified yet."

This isn't a technical problem. It's an architecture problem. The old framework assumed scarcity of output. The new reality is abundance of output and scarcity of judgment.

Four Questions Your System Should Answer

If your productivity system still assumes you're single-threaded, you're managing blind. Here's the test I'm using with clients — four questions that expose whether your framework has caught up to how you're actually working:

1. How many threads are running right now — exact number?

Not "a few things in progress." Exact count. If you can't answer this, you don't have visibility into your actual workload. I keep a live doc. Currently: six threads. Three research agents, one code review running overnight, two client deliverables in draft with collaborators.

2. Which ones are blocked specifically on a decision from you?

This is your real to-do list now. Not "write the report" — the agent's writing it. Your task is "decide whether the report should emphasize regulatory risk or operational risk." That decision is the only thing preventing the thread from completing.

3. Which ones can finish without ever touching your inbox again?

If you can't answer this, you're going to become the bottleneck on work that doesn't need you. I have two threads running right now that will complete, get filed, and never require my review. The agent knows the success criteria. I only need to know if something fails.

4. What's your cadence for reviewing the ones that come back?

If you don't have a rhythm for this, you're going to drown in completed work. I do output review twice a day now — mid-morning and end of day. Anything that comes back outside those windows waits. Otherwise I'm constantly context-switching to review things that could batch.

The Uncomfortable Middle

Here's what I don't know yet: what happens when the number of supervisable threads exceeds human judgment capacity?

Right now I can track six threads. Maybe ten on a good day. What happens when the bottleneck isn't my ability to make decisions, but my ability to even know which decisions are waiting? We're optimizing for parallel execution while our brains are still fundamentally serial.

I don't have a clean answer. I've watched three technology cycles disrupt work — internet, mobile, cloud. Each time, the people who survived weren't the ones who predicted the future. They were the ones who noticed when their tools stopped fitting the problem.

Your to-do list still matters. The single-threaded framework around it doesn't.

What to Do Monday Morning

Open your task manager. Whatever you're using — Todoist, Things, Notion, a legal pad.

Count how many items on that list are actually running right now without you. Not "planned" or "someday." Running. Being executed by an agent, a collaborator, an automated workflow.

If the answer is zero, you're still operating single-threaded. That's fine — until it isn't.

If the answer is more than three, ask yourself: where are you tracking what's blocked on you versus what's just in progress? If those are mixed together in one list, you've already lost visibility.

I'm not saying abandon GTD. I'm saying the "next action" paradigm assumes you're the action. When five things are already in motion, the next action is deciding which thread gets your judgment next.

The railroad didn't make towns obsolete. It just made the ones far from the tracks irrelevant. Same pattern here. Nobody's getting fired the day they start using agents. But the gap between people who learned to manage threads and people still managing tasks? That gap is opening faster than any technology shift I've seen.

What's running right now that you're not tracking?