AI + GTD: How Artificial Intelligence Upgrades the Getting Things Done System

GTD's core insight is still valid — but AI handles the capture and review phases automatically. Here's how pairing AI with the Getting Things Done framework eliminates the friction that causes most people to abandon it.

David Allen published Getting Things Done in 2001. The book sold millions of copies, spawned entire communities of productivity enthusiasts, and introduced terms like "trusted system" and "next action" into the lexicon of knowledge workers worldwide. And yet — most people who read it don't actually stick with GTD for more than a few weeks.

The reason isn't that the system is wrong. It's that it requires constant manual upkeep. Capturing every open loop into an inbox, doing weekly reviews, maintaining project lists, processing every item into next actions — all of that overhead is real work that piles on top of your already full day.

AI doesn't invalidate GTD. It removes the parts that were always friction points and lets you focus on the parts that require genuine human judgment.

A quick refresher: the 5 steps of GTD

Before getting into how AI fits, it helps to have the framework clear. GTD organizes work into five stages:

1. Capture — Get every commitment, idea, and open loop out of your head and into a trusted external system.
2. Clarify — Process each item: Is it actionable? What's the next physical action? Does it belong to a project?
3. Organize — Sort items into the right buckets: next actions, projects, waiting for, someday/maybe, reference.
4. Reflect — Review your lists regularly (especially the weekly review) to stay current and trust your system.
5. Engage — Choose what to work on with confidence, knowing your system is complete and current.

The promise of GTD is that a complete, trusted system frees your mind. You don't carry open loops in your head because everything lives somewhere reliable. The problem has always been building and maintaining that completeness.

Step 1: AI solves the capture problem

Capture is the most important step and the one people do worst. Allen famously said your inbox should be the only place commitments live — but in practice, commitments arrive in five different apps, your phone's voice memos, Post-it notes, and the thing someone mentioned to you in the hallway.

The GTD practitioner's capture problem isn't motivation — it's surface area. No one can realistically funnel every Gmail thread, every Notion comment, every calendar invite, and every Slack ping into a single inbox by hand.

This is exactly what an AI layer can do. A tool like REM Labs connects directly to Gmail, Google Calendar, and Notion and reads the last 90 days of activity across all of them. Rather than requiring you to manually capture every incoming signal, the AI monitors those surfaces automatically and extracts what matters — commitments people made to you, deadlines embedded in email threads, open questions that need a decision.

The result: your capture phase happens passively. You wake up and the things that arrived overnight are already captured. You didn't have to do anything.

Step 2: AI-assisted clarification — flagging what needs a decision

Clarify is where most GTD systems fall apart during the weekly processing session. You look at a captured item — "email from Marcus about the proposal" — and now you have to decide: Is this actionable? What's the next action? Is it waiting for someone? Does it belong to a project?

Done honestly, this takes real mental energy per item. When you have 200 items in your inbox after a week of travel, the cognitive load of clarifying each one is punishing enough that people just declare inbox bankruptcy and start over.

AI can pre-clarify many items before you even look at them. It can read an email thread and determine: this is a request that needs a response, this is information to file, this is a commitment someone made to you that you should track. It won't get every item right — but it reduces your clarification queue from 200 ambiguous items to 20 items that genuinely need human judgment.

The AI handles the easy determinations. You handle the ones that require context only you have: whether a project is still a priority, whether a relationship warrants a follow-up, whether an opportunity is worth pursuing.

Step 3: Organize — where structure still comes from you

AI can suggest where things belong, but the organizational structure of your GTD system is inherently personal. How you define a "project" (is "prepare for board meeting" one project or five?), what belongs in someday/maybe versus active next actions, how you tag by context — these decisions reflect your values and your work, not patterns an AI can fully infer.

The practical division here: let AI populate and suggest, but own the structure yourself. If AI surfaces "follow up with Sarah on budget proposal" as an action item, you decide whether that belongs under the client project, your finance review project, or a waiting-for list. That judgment call takes three seconds when AI has already done the extraction.

Step 4: AI replaces the weekly review

The weekly review is the heartbeat of GTD. Allen prescribes two hours every Friday to review every project, every next action list, every calendar event from the past week and the coming two weeks, every open loop. It's comprehensive. It's also the first thing people skip when they get busy — which is exactly when they need it most.

This is where AI creates the biggest structural improvement to GTD. Rather than requiring a dedicated two-hour block to manually survey your system, an AI-powered morning brief performs a version of that review every single day — automatically.

REM Labs does this by reading across your Gmail, Notion, and Calendar data overnight and delivering a morning brief that surfaces: what's due soon, which conversations have gone quiet that shouldn't have, which calendar blocks suggest you're overcommitted in a given area, and which projects have had no activity recently. That's essentially the core of a GTD weekly review, delivered daily, without requiring a Friday afternoon you probably don't have.

Step 5: Engage with confidence

The goal of steps one through four is to reach step five in the right mental state. Allen calls it the "mind like water" state — you can respond to whatever arrives without feeling like you're dropping something else. You engage with your current task fully because you trust your system has everything else covered.

When AI handles capture and review, this state becomes easier to access. You start the day with a brief that tells you what actually matters today. You don't spend the first hour triaging email trying to reconstruct your priorities. You engage from a position of awareness rather than reaction.

Where you still need human judgment

AI-enhanced GTD is not AI-automated GTD. Several parts of the system still require you:

• Deciding what a "project" is. Only you know your goals well enough to define your project landscape correctly.
• Setting priorities across projects. AI can surface urgency signals, but the question of which project matters most to your long-term goals is a values judgment.
• Closing loops that require conversation. AI can remind you to follow up with someone; it cannot have the conversation for you.
• Saying no. AI can show you that your calendar is already full. It can't decline commitments on your behalf or negotiate scope.
• Deciding what goes in someday/maybe vs. the trash. That distinction is a values judgment about what you actually want.

Think of AI as handling the administrative skeleton of GTD — the collecting, the reviewing, the pattern-spotting — while you handle the decisions that require judgment, relationships, and knowledge of your own goals.

A practical AI-enhanced GTD setup

Here's what a working AI + GTD setup looks like in practice:

Your capture system

Connect your primary communication channels to an AI layer (Gmail, Calendar, Notion at minimum). Let the AI monitor those surfaces and flag items that need attention. Keep a quick-capture tool — a phone note, a voice memo, a notebook — for things that don't arrive digitally. Process that physical inbox once a day into your system.

Your daily review

Replace the weekly review with a daily morning brief from your AI. Read it at the start of your day. It should tell you: what's urgent today, what conversations are stalling, what's coming up in the next few days. Use it to set your top three priorities for the day.

Your weekly check-in

You still want a weekly moment — but it's shorter now. Spend 20–30 minutes on Fridays doing what AI can't: reviewing your project list against your goals, moving items from someday/maybe to active (or deleting them), and deciding what matters most in the coming week. Let that decision inform how your AI brief is weighted.

Your project lists

Maintain these in Notion or whatever project tool you use. If your AI layer connects to Notion, it can surface project activity patterns ("this project hasn't had any Notion edits in two weeks") that keep your project list honest.

The bottom line on AI GTD productivity

GTD's fundamental insight — that your mind is for having ideas, not holding them — is more relevant than ever. The volume of inputs knowledge workers deal with in 2026 has increased dramatically since 2001. Email, Slack, Notion, GitHub, calendars, voice messages — the surface area of open loops is enormous.

AI doesn't replace GTD. It makes GTD's promise finally achievable at scale. Capture happens automatically. Review happens daily without a dedicated two-hour block. You spend your GTD energy on the parts that only humans can do: deciding what matters, building relationships, choosing what to work on next.

The system finally trusts itself — because it's always being updated.