A task list tells you what you owe. It does not tell you whether today can hold it. That is a different question, and unlike most questions about work, it has an arithmetic answer.
Auto-schedule
One action in the planner takes your open tasks and places them into the gaps between today's meetings.
What it does, in order
It reads your day: your working hours, and every meeting already on your calendar.
It works out the free gaps. Not the total free hours, which is a misleading number, but the actual contiguous blocks left between commitments. Four separate twenty-minute gaps are not an eighty-minute block and treating them as one is how optimistic plans get made.
It takes your open tasks that carry an estimate and are not already pinned to a specific time.
It sorts them by priority first and by length second, so the urgent things get placed before the convenient ones.
It fills the gaps from the front and shows you the result.
Nothing is committed until you accept it.
The estimate is the whole thing
A task with no estimate is given a default half hour. That default is a courtesy rather than a guess at your actual work: half an hour is what an unestimated task is worth to an algorithm, and it will frequently be wrong.
Add estimates while you capture, in the same sentence. "Review the contract for 45 minutes" costs nothing extra to type and makes the difference between a schedule and a suggestion.
What it deliberately ignores
Blocks already derived from tasks are not treated as meetings when it calculates your gaps, so running it twice does not schedule around its own previous attempt and slowly fill your day with nothing.
What it is actually for
The value is not the schedule. The schedule will not survive contact with the morning, and everybody knows it.
The value is the answer to whether today is possible, available at nine in the morning rather than at four in the afternoon.
Four hours of tasks and five hours of meetings is a fact. Having it early gives you three honest responses:
Move something. Most weeks, one thing genuinely can be tomorrow, and choosing which is easier before the day starts.
Shorten a meeting. An hour that could be thirty minutes usually can be, and the person who called it will rarely object.
Tell somebody now. The cost of saying "this will be Thursday" at nine in the morning is close to nothing. The cost of saying it at six in the evening is a broken commitment and a colleague who has spent the day waiting.
Protecting the result
An accepted schedule puts blocks on your calendar, which means colleagues looking for a slot can see them.
That is most of the benefit, and it is easy to miss. Time that exists only inside a task list is invisible to everybody else, and invisible time gets booked over by people acting entirely reasonably on the information they had.
The rhythm that works
Capture with estimates all week. Two extra words per task, done at the moment you already have the context.
Plan once, in the morning. Five minutes, before the first meeting.
Re-plan when the day genuinely breaks, not continuously. A schedule you rebuild four times is a schedule you are hiding in, and rebuilding it feels productive in a way that it is not.
Look at what never fits. Over a month, the tasks that are never scheduled are the most useful list in the system. Either they do not actually matter, in which case delete them, or something structural is wrong with the shape of your week, in which case the task list was never the problem.
Estimating without agonising
The objection to estimates is that they are always wrong. They are, and it does not matter, because the scheduler needs an order of magnitude rather than a prediction.
Three bands cover almost everything.
Fifteen minutes. A reply, a review, a decision, a call you have to make.
Half an hour. The default, and the right answer for most discrete pieces of work.
An hour or more. Anything requiring you to load a subject into your head before you can start.
If a task genuinely needs more than two hours, it is not a task. Break it into subtasks and estimate those, because a three-hour block will never find a gap and the task will sit unscheduled indefinitely while you wonder why.
Estimate at capture, not later. You have the context in your head at the moment you write the task and you will not have it on Monday morning.
Reading the result honestly
The schedule it produces is a proposal, and there are three sensible reactions.
It fits with room to spare. Accept it and stop planning. The temptation to fill the remaining gaps is how a comfortable day becomes a full one.
It fits exactly. Treat that as full, not as achievable. A day with no slack breaks on the first unexpected thing, and there is always one.
It does not fit. This is the useful outcome and the reason to run it. Something moves, something shortens, or somebody hears about it now rather than at six in the evening.
What it will not do
It will not protect your day. Blocks are visible, not immovable, and a colleague can book over one. That is correct behaviour: the alternative is a calendar nobody else can use, which makes you the bottleneck you were trying not to be.
It will not learn your pace. It uses the estimate you gave it. If you consistently take twice as long as you predict, the fix is better estimates rather than a cleverer scheduler, and noticing the pattern is most of the fix.
It will not schedule what it cannot see. A task with no estimate and no due date is invisible to it. That is by far the most common reason the result looks emptier than the week feels, and it is a two-minute fix.
It will not decide what matters. Priority is your input. An auto-scheduler fed a list where everything is urgent produces a schedule in the order the tasks happened to be created.
Did this answer your question?
No, ask a person