Fix the bottleneck. Keep the system small.

1. Capture every open commitment without ranking it yet.
2. Choose the one outcome that most changes the week and define its next physical action.
3. Limit active outcomes to three; park everything else in a visible queue.

A visible limit forces trade-offs and stops every incoming request from becoming active work.

1. Before ending today, choose tomorrow's must-win result.
2. Put a 60–90 minute block on the calendar with a specific finish line.
3. Start with the source file open and communication tools closed.

A scheduled cue turns a vague intention into a concrete commitment and exposes capacity conflicts early.

1. Rewrite the task as a visible action: “Open the brief and draft the three section headings.”
2. Create a cue: “At 9:00, after coffee, I will open the brief at my desk.”
3. Commit only to the first five minutes; continue if momentum arrives.

Specific cues and small entry actions reduce ambiguity and the activation cost of beginning.

1. Turn off non-human and non-urgent notifications at the operating-system level.
2. Publish one emergency channel people can use during focus blocks.
3. Use a blocker only for sites or apps you still open reflexively.

Fewer external cues mean fewer forced task switches and less time rebuilding the working context.

1. Choose one visible “doing” item and keep the rest queued.
2. Group calls, reviews, admin, and writing into separate modes.
3. Before a necessary switch, write the exact next step and unresolved question.

Task switching carries cognitive setup costs; resumption notes reduce the cost when switching is unavoidable.

1. Describe the user-visible or client-visible finish line.
2. Break the path into demonstrable milestones with one owner each.
3. Compare estimate ranges with actuals and adjust before the next commitment.

Visible scope, ownership, and uncertainty reveal delivery risks before the deadline does.

1. Capture the follow-up when the request is sent, not when it becomes overdue.
2. Record who owns the next move and the date you will check again.
3. Sweep the waiting list daily for urgent work and fully during the weekly review.

Externalizing commitments removes memory as the control system and makes stalled work visible.

1. Replace status meetings with a short written or recorded update.
2. For necessary meetings, publish the decision needed and pre-read in advance.
3. End with the decision, owner, deadline, and written record.

Asynchronous status sharing protects shared focus time; explicit outcomes stop meetings from creating more meetings.

1. Choose two or three response windows that fit the service level your role requires.
2. Create VIP and true-urgency exceptions; mute everything else between windows.
3. Use filters and templates for repeatable requests, then close the inbox.

Batching reduces repeated checking and protects attention while preserving a clear response promise.

1. Name one directly responsible owner for the current stage.
2. Attach the source, decision history, constraints, and definition of done.
3. Ask the receiver to confirm the next action or clarify the gap immediately.

Clear interfaces reduce ambiguity precisely where context is most likely to be lost.

1. Run the process manually while recording every step and exception.
2. Template the stable inputs and outputs; remove unnecessary steps.
3. Automate the reliable remainder and add a failure notification.

Standardization removes variation before automation multiplies it.

1. Create one canonical project page and link existing sources rather than copying them.
2. Log decisions with date, owner, rationale, and what would reopen the choice.
3. Keep the current status and next action at the top; archive stale detail below.

A stable retrieval path reduces search time and makes project continuity independent of memory.

1. After learning, close the source and reconstruct the key idea from memory.
2. Create a small set of questions that return at increasing intervals.
3. Apply the concept in a realistic task and explain the result in plain language.

Retrieving knowledge strengthens later access more effectively than passive rereading alone.

1. Choose a normal shutdown time and define the emergency exception explicitly.
2. Place brief movement and meal breaks between demanding blocks.
3. Reduce committed work when energy, errors, or sleep trend in the wrong direction.

Sleep and recovery support attention, memory, judgment, and the ability to sustain performance.

1. Write an if-then plan tied to a cue that already occurs reliably.
2. Make the minimum version small enough to complete on a low-energy day.
3. Prepare the environment in advance and review completion by cue, not by streak pride.

Specific implementation intentions help translate a strong goal into action when the cue appears.

1. Identify repeated choices that have a safe, reversible default.
2. Document the default, exception threshold, and who can decide.
3. Log only consequential or hard-to-reverse decisions with rationale.

Defaults remove repeat deliberation while exception rules preserve judgment where it matters.

1. Choose the canonical home for tasks, time, and project reference.
2. Delete duplicate capture points and stop maintaining unused metadata.
3. Cap system administration and keep only fields that change a decision.

A smaller system reduces tool switching, reconciliation, and the temptation to optimize appearances over output.

1. Define the deliverable, audience, constraints, and what the model must not assume.
2. Provide approved sources or examples and ask the model to expose uncertainty.
3. Verify facts, logic, privacy, and final fit before publishing or acting.

AI is most useful as a bounded accelerator; a verification gate keeps speed from becoming downstream defects.