Plan a Workday That Fits—and Replan When Time Disappears

To make a feasible day plan, subtract fixed commitments from the work window, then allocate the full estimated duration of each task according to an explicit selection rule. Check continuous blocks as well as total minutes. Keep every task that cannot fit in an unscheduled list, and rebuild the whole plan when availability changes.

A packed calendar can conceal an impossible day. A 90-minute task does not fit four separate 15-minute gaps, and a meeting added to the afternoon can push work out even when the morning stays unchanged. The useful output is a full timetable plus the work it leaves out.

This guide works through one complete original fictional day: seven clarified tasks, a first plan, a new fixed commitment and a full replacement plan. The inputs, wrong examples and repairs are authored teaching material, not captured assistant output or a record of completed work. The arithmetic was checked locally; it does not measure how accurately anyone can estimate a real task.

Start once the inputs are decided

This task starts after task clarification. You need to know who is doing the work, what each task includes, the person's accepted estimate, dependencies, deadlines, split permissions and fixed availability. An assistant cannot derive a reliable duration or deadline merely from a vague note. If those facts are still undecided, resolve them before treating a timetable as feasible.

For this example, Rowan has made those decisions. Here is the full input, also available as a separate case file.

This is original fiction. Rowan is the fictional planner and sole worker; every task below belongs to Rowan. The tasks are already clarified. The durations are estimates Rowan explicitly accepts for this exercise, not measured completion times or promises. No work has started and no task is reported complete.

Day and fixed commitments

• Plan date: 14 May 2030. Every time in this packet is UTC.
• Work window: 14 May 2030, 09:00–17:00 UTC, with no overtime.
• Both V1 and the later V2 are prepared before work starts. V1 is prepared at 08:30 UTC; the one change arrives at 08:45 UTC. These are fictional scenario times, not actual execution timestamps.
• One person can do only one activity at a time. Start is inclusive and end exclusive: adjacent blocks may share an endpoint.
• All starting source materials are available at 09:00. T02 still requires the exception list produced by T01; the only task dependency is T01 → T02. No travel, setup or waiting time is required beyond the estimates and fixed blocks explicitly supplied here. This simplified assumption belongs to the example; add those costs when they apply to your day.
• Do not move, shorten or use fixed meetings or breaks for tasks. Preserve every task and every estimate in both versions. Do not infer completion from a block's clock time.

Every task and its accepted estimate

Every deadline is an inclusive latest finish, not an intended start. T07's next-day deadline does not authorize working outside today's window or invent a booking for tomorrow.

Rowan's explicit selection rule

Rowan accepts the priority order 1, 2, 3, 4, with 1 highest. Within a priority, use the earlier deadline; if deadlines tie, use the lower numeric task ID. The resulting order is T01, T02, T03, T04, T05, T06, T07.

Reserve fixed commitments first. Consider tasks in that order, retaining earlier reservations. Place the whole task at the earliest legal time that fits its duration, dependency and deadline. For T03, use the earliest legal allocation of all 90 minutes in one or two blocks, with each block at least 30 minutes. Never place only part of a task and call it scheduled. If the task cannot fit, leave it unscheduled with its full estimate and continue to the next task; a lower-priority task may use a gap that cannot hold the higher-priority one. Do not move earlier reservations merely to fit a later task. This is Rowan's chosen planning rule, not a claim of globally optimal scheduling.

Report the whole day, including unallocated time. For each of the seven IDs, report scheduled and unscheduled minutes and a reason. Unscheduled means not allocated in this plan, not canceled, completed or booked for another day. Only Rowan may change estimates, priorities or deadlines, permit a different split, or agree to a new commitment. Make those changes explicit before replanning.

Compare the requested work with the time available

The working window contains eight hours, or 480 minutes. The original fixed commitments use 120 minutes, leaving 360 for tasks. Rowan's estimates total 450 minutes. Even before placing a single block, at least 90 task minutes cannot be allocated under these inputs.

That lower bound is not the final unscheduled total. Tasks have shapes: most need one continuous block, and T03 can use only one or two blocks of at least 30 minutes each. After higher-priority reservations, smaller gaps may remain unusable for the remaining work.

Read the whole first plan and its replacement

The following tables are the complete corrected work product. The separate plan file contains the same two versions for comparison. U1, the only change, arrives at 08:45 UTC on 14 May 2030, before work starts: add an immovable equipment walkthrough at 13:00–14:00 UTC. All other inputs stay the same. The full change note preserves this timing explicitly.

V1: before the extra commitment

V1 task disposition

V1 arithmetic: 480-minute window − (30 + 45 + 30 + 15) fixed = 360 available task minutes. Demand is 45 + 75 + 90 + 60 + 90 + 30 + 60 = 450 minutes. Tasks receive 45 + 75 + 90 + 60 + 30 = 300 minutes. Four 15-minute gaps total 60 minutes. Thus 120 fixed + 300 task + 60 unallocated = 480, and 300 scheduled + 150 unscheduled = 450.

The demand exceeds capacity by 90 minutes even before considering block shape. The actual unscheduled total is 150 minutes because the remaining 60 minutes are four separate 15-minute gaps. They cannot satisfy either unsplittable task. T05 is considered before T06; its higher priority does not make its 90 minutes fit. No deadline is moved or silently waived: T05 remains due today and unallocated, requiring Rowan to resolve the shortfall.

V2: the complete replacement after U1

F05 is added at 13:00–14:00. The update arrives at 08:45, so the whole day is still a plan. Original fixed blocks remain at their original times.

V2 task disposition

V2 arithmetic: 480 − (120 original fixed + 60 new fixed) = 300 available task minutes. Demand stays 450. Allocations are 45 + 75 + (30 + 60) + 30 = 240 minutes, with 60 unallocated. Thus 180 fixed + 240 task + 60 unallocated = 480, and 240 scheduled + 210 unscheduled = 450. Demand now exceeds capacity by 150 minutes; block constraints leave another 60 minutes unused.

What changed, and what still needs a decision

T04 and T05 are still due on 14 May. Rowan needs to choose a scope, priority or deadline change, or obtain an explicit agreement to move that work. This packet makes none of those decisions. If either estimate proves wrong in practice, record what actually happened and replan from that evidence; the mathematical fit does not promise completion.

Repair the full day, not just the overlapping block

The complete authored wrong versions deliberately show two ways to produce a convincing but unusable timetable. WRONG-V1 gives every task its full duration, yet overlaps work with meetings and puts T02 before its prerequisite. WRONG-V2 includes the new commitment but compresses two tasks, loses T07 and extends T06 beyond the day. Both require whole-plan replacement.

Use the complete repair request to inspect the same task with all original inputs, the change and both wrong tables included. It contains no corrected answer. If you give it to an assistant, ask for both full versions and every task's disposition back. A list of detected errors alone is not a usable replacement day.

The author's arithmetic record reports actual local calculations on these tables: row durations, overlaps, fixed blocks, task totals, split limits, dependencies, deadlines and minute conservation. It also records the deliberate faults in the wrong tables. This is an author check, not an independent review or a scheduling-product test.

Make the remaining decision visible

The unused 60 minutes are four 15-minute gaps, not an invitation to shrink a 60- or 90-minute task. Under Rowan's rule, V2 retains T03 at its full estimate, leaves T04 and T05 unallocated even though they are due today, and uses a suitable slot for T06. That is a planning shortfall Rowan must act on; the calendar has not resolved it.

Rowan could explicitly revise scope, agree a later deadline, change priority or authorize another split. Each would be a new input. Rebuild the plan using that decision rather than silently rewriting the original estimate or marking the task done. When work has actually started, keep a separate execution record: what finished, what remains and when a new commitment arrived. This example's before-work update does not demonstrate replanning around partially completed work.

You can do the same arithmetic with paper time blocks or a calendar you edit by hand. For a short day, that may be less work than explaining the constraints to an assistant. A priority queue without exact start times can also suit work whose durations are too uncertain for a tight timetable; it answers what to attempt next rather than proving a whole day fits.

A public weekly-planning discussion describes using an assistant to clarify work and reserve time. A separate request for a non-AI time-blocking app is a useful reminder that AI is optional. These are limited individual accounts; the capacity and replan exercise here is our editorial design, not a result established by those posts.

If you dictate the planning input, check task IDs, numbers, time zone, deadlines and words such as “cannot split” against the transcript. Typing is equally suitable. The useful standard is the same: every minute and every task has a visible place, including the work that does not fit.