Most frameworks for deep work tell you what to protect. Few tell you what to build.
Protecting a calendar block is necessary. But protection is passive. What happens in the minutes before a session begins—the cognitive preparation, the interruption resolution, the goal-setting—determines whether that block produces depth or produces the appearance of depth.
We built the Deep Work Runway to solve the entry problem. Here is the full framework.
Why Deep Work Needs a Runway
Newport’s definition in Deep Work is precise: “professional activity performed in a state of distraction-free concentration that pushes your cognitive abilities to their limit.” The phrase “state of” is the important part. It describes something you enter, not something that automatically exists when you sit at your desk.
Pilots do not take off without a runway. The runway is not wasted distance—it is the acceleration phase that makes flight possible. Knowledge workers need the equivalent.
The cognitive obstacles to entering depth are well-documented:
Attention residue (Sophie Leroy, 2009): Cognitive threads from previous tasks persist into new ones, degrading performance on work requiring full engagement. The half-finished email you left, the meeting you just walked out of, the message you intended to respond to—all of these leave residue.
Context initialization cost: Loading the relevant mental model for a complex task takes time and working memory. If you are doing that initialization while simultaneously trying to work, both suffer.
Goal ambiguity: Csikszentmihalyi’s research on flow states identifies clear goals as a prerequisite. Sessions that begin with a vague objective tend to produce unfocused work, because the mind cannot enter deep engagement without something concrete to aim at.
The runway addresses all three, gate by gate.
Gate 1: Prime Context
What this gate solves
When you sit down to resume work, your working memory does not automatically reload. Research by Cowan suggests working memory holds roughly four chunks of information at a time. A complex project might require holding: the current state of the argument, the three strongest counterpoints, the gap in the data, and the structure of the section you are building. None of that is in memory when you arrive.
The standard workaround is re-reading—going back through your notes, your previous draft, the relevant documents until you remember where you were. This takes fifteen to thirty minutes and is cognitively passive. You are not working; you are initializing.
What AI does here
You provide the raw material—notes, a previous draft, a task description, relevant emails—and ask AI for a condensed working-memory brief.
The prompt structure:
I am beginning a session on [project/task].
Here is my relevant material:
[paste notes, last draft section, or describe the current state]
Produce a working-memory brief:
1. Two sentences on where I left off.
2. The single highest-leverage next action.
3. Three things I need to hold in mind to do it well.
4. One risk or complication to watch for.The output is not a plan. It is a cognitive primer—the minimum information needed to start from a position of orientation rather than confusion.
What good Gate 1 output looks like
Good Gate 1 output is short. Three to five sentences. It should orient you without overwhelming you. If you are reading a paragraph, ask AI to cut it by half. If you cannot identify the next action from the brief, the brief is too abstract.
Gate 2: Kill Interruptions
What this gate solves
Attention residue is not only caused by previous tasks. It is also caused by future tasks you are worried about. The message you have not replied to. The colleague who might need you. The meeting that starts ninety minutes into your session.
Gloria Mark’s research on attention switching found that even when workers are not interrupted, the anticipation of interruption degrades focus quality. You are cognitively half-present because part of your mind is monitoring for the things that might need you.
This is not a character flaw. It is rational vigilance in an environment that has trained you to be interruptible. The solution is not to force yourself to stop monitoring. It is to actually resolve the things being monitored.
What AI does here
List every open loop that might surface during the session. Give AI the list and ask for resolution strategies.
I am going offline for [X] minutes starting now.
These open loops might pull my attention during the session:
[list them—even the trivial ones]
For each one, give me:
- A 30-second action to close it right now, OR
- A brief message that creates a clean handoff, OR
- Confirmation it can genuinely wait until after.Work through the list before you close the AI. Send the quick messages. Resolve the two-minute items. Confirm that the rest can wait.
The mental mechanism
The value here is not just triage—it is closure. Leroy’s research shows that attention residue decreases when tasks are either completed or cleanly handed off. A message that says “on this after 3pm” creates cognitive closure on the open loop. Your mind releases it from active monitoring.
Five minutes of triage at Gate 2 creates ninety minutes of quiet.
Gate 3: Set the Exit Point
What this gate solves
Sessions without exit points end in one of three ways: you work until time runs out (exhausted, unclear whether you succeeded), you drift until you stop caring (unfocused output, no sense of accomplishment), or you extend endlessly (which damages the next day’s sessions by depleting cognitive resources).
The problem is not a lack of ambition. It is the absence of a concrete finish line.
What AI does here
Give AI your task and your available time. Ask for a session-specific, observable deliverable.
My session goal is: [describe the task or project area]
I have [X] minutes.
Define a specific, observable exit point for this session.
It should be:
- Ambitious but achievable in the time I have
- Observable—I can look at it and say "yes, done" or "no, not yet"
- Narrower than my overall project goal
Not "make progress"—an actual concrete output.Write the exit point somewhere visible. A sticky note works.
Why observable specificity matters
The difference between “work on the report” and “complete the executive summary and draft the first two findings” is not just precision. It changes how you work. With a concrete exit point, you make decisions during the session based on whether they advance you toward that specific output. Without one, every decision is up for grabs.
Csikszentmihalyi found that clear goals are one of the most consistent predictors of flow states. The exit point is that goal—specific to this session, not to the project as a whole.
The Gates in Sequence
The three gates have a deliberate order.
Gate 1 (context) comes first because you cannot assess your interruption landscape or define a realistic exit point without knowing where you are. If you do not know where the work stands, you cannot triage what might pull you from it or decide what success looks like.
Gate 2 (interruptions) comes second because unresolved loops will contaminate the exit-point definition. If you are anxious about an open item, you will set a shorter or vaguer exit point to leave yourself room to handle it.
Gate 3 (exit point) comes last because it requires the combination of context (what is the work) and cleared anxiety (what can I commit to) to be calibrated accurately.
Run them in order. It takes five to eight minutes total.
After All Three Gates: The Hard Rule
Close AI. Not minimize—close.
The runway is preparation. The session is work. They are distinct. If AI is visible, you will query it. Querying it is a context switch. Context switches generate attention residue. Attention residue degrades the depth you just spent five minutes building.
The runway earns the session. The session is the point.
Tracking and Improving the Runway
After each session, spend two minutes on a brief review:
- Did the Gate 1 brief accurately orient me?
- Did Gate 2 actually quiet the open loops, or did something still pull my attention?
- Did I hit the Gate 3 exit point?
- What was the first distraction, if any?
After a few weeks of this data, you will see patterns. Some project types require longer Gate 1 input. Some interruption sources are chronic and need systemic solutions rather than session-by-session triage. Some exit points are consistently too ambitious or too conservative.
Tools like Beyond Time can make this review faster by keeping your session log alongside your planning context—so you are reviewing actual patterns rather than reconstructing from memory.
The Runway Is Not a Silver Bullet
Two things the runway cannot fix:
An environment that makes depth structurally impossible. If your organization expects instant response, values availability over output, and schedules meetings in every open slot, the runway helps at the margins but does not solve the underlying problem. That is an organizational issue requiring a different intervention.
Sustained practice deficit. Newport is explicit that deep work capacity is built over time. If you have spent years working in fragmented, interrupted conditions, your tolerance for sustained concentration will be low. The runway helps you enter depth more reliably, but it does not replace the work of building the capacity through repetition.
For the right problems—knowledge workers who have the time blocked but consistently fail to enter depth—the runway is highly effective. It solves a specific, tractable problem: the entry cost.
Start Here
Before your next session, run only Gate 3. Paste your task into AI and ask it to define a specific, observable exit point. Then write that exit point somewhere visible and start working.
You will feel the difference immediately.
Related:
- The Complete Guide to Deep Work with AI Assistance
- The Complete Guide to Deep Work Scheduling with AI
- The Complete Guide to Time Blocking with AI
Tags: deep work, framework, AI tools, focus, attention residue
Frequently Asked Questions
-
What is the Deep Work Runway framework?
The Deep Work Runway is a three-gate pre-flight process that uses AI to prepare the conditions for deep concentration. Gate 1 primes working-memory context. Gate 2 eliminates interruption anxiety. Gate 3 sets a concrete exit point. All three gates complete before the session begins and AI is closed.
-
How is the runway different from a standard pre-work routine?
Most pre-work routines are behavioral—a cup of tea, putting on headphones, opening a specific app. The runway is cognitive. It uses AI to resolve specific mental bottlenecks: missing context, unresolved open loops, and ambiguous success criteria. It is faster and more targeted than a ritual alone.
-
Can the Deep Work Runway work without AI?
Yes, with modifications. You can prime context manually by reviewing notes, triage interruptions with a written list, and define exit points on a notepad. AI makes Gate 1 faster and Gate 2 more thorough—but the three-gate structure itself is valid regardless of tooling.
-
How long does running the runway take?
Five to eight minutes for experienced practitioners. When you first start, expect eight to twelve minutes. The process speeds up as you learn what level of detail produces the right output at each gate.