Teachomatic what to automate, and what not to

What Actually Saves Time

The claim that AI saves teachers time is the best-supported claim in this whole field, and it is still routinely overstated. The research puts it at around half an hour a week in one study that measured it properly, without a drop in quality.

Half an hour. Not an afternoon.

For a workplace-oriented comparison outside education, this discussion offers another way to look at measurement, workload, behaviour, or accountability.

That is worth having. It is also worth knowing before you reorganise anything, because the difference between the real figure and the advertised one is where disappointment comes from — and disappointment is what makes teachers abandon a tool that was genuinely helping a bit.

This page is about where that half hour actually comes from.

For an external perspective on lesson planning, teacher workload, and classroom technology, see K-12 Dive.

Where the time really comes back

The pattern across the tasks that work is consistent: you save time when the hard part is starting, not deciding.

First drafts of things you will rewrite anyway. A parents' evening summary, a scheme-of-work outline, a set of comprehension questions. The blank page costs more than the editing does, and a mediocre draft removes the blank page.

Reformatting and reshaping material you already have. Turning a passage into three reading levels, a text into a cloze exercise, prose into a table. The thinking is done; the labour is mechanical, and levelling a text for a mixed class is the clearest case on the whole site.

Varying something that exists. You have one good worksheet and need six more questions in the same shape. This is the highest-yield category on the list and the least glamorous.

Explaining the same idea five different ways. Genuinely useful when a class has not understood and you have run out of your own analogies.

Administrative prose. The email to the department, the risk assessment paragraph, the paperwork that has to exist and that nobody reads carefully. This is the least discussed and possibly the largest single saving.

Where it does not

Anything requiring knowledge of the particular child. A comment on a report about a specific student, a decision about who to seat where, a judgement about whether someone is struggling or coasting. The output looks right and is generic, and generic here is worse than nothing because it reads as if you were not paying attention.

Marking that is actually assessment. Working out what a class has and has not understood is not clerical work with a slow interface. It is the diagnosis, and it is most of what makes the next lesson good.

Anything you would have to check line by line anyway. If verifying the output costs as much as producing it, you have moved the work rather than removed it. This is the trap in anything factual: dates, formulae, citations, historical detail, anything where a plausible wrong answer is worse than no answer.

Tasks you only do twice a year. The setup and the learning cost more than the task.

Anything where the doing is the thinking. Some planning is how you work out what the lesson is. Delegating the artefact can quietly delegate the thinking, and you find out in the lesson.

The catch nobody advertises

Studies keep finding the same thing: teachers do not bank the saved time. They spend it on more or better feedback.

That is probably a good outcome for students. It is not the outcome that was promised, and it is worth deciding deliberately rather than drifting into it.

So decide in advance what a saved hour is for. More detailed feedback is a legitimate answer. So is finishing at a reasonable hour. What does not work is leaving it undecided, because undecided time gets absorbed by whatever expands to fill it, and then you have adopted a new tool, learned it, and cannot say what you got.

Measure it for two weeks

The honest way to find out whether any of this is true for you, and it costs almost nothing.

Pick two tasks from the first list. Not five. Time them by hand for two weeks before, then two weeks using the tool, and write down the numbers.

Also record two things that are not minutes. How good was the output before you edited it — usable, needed a rewrite, unusable. And did you use the saved time on purpose.

At the end you will know more about your own practice than any vendor's case study can tell you, and you will have a number you can give a colleague who asks.

What to do first, if you do one thing

Take the single administrative writing task you resent most — the recurring email, the standard paragraph, the form that has to be filled the same way every term.

Do that one. It is low stakes, it is checkable in seconds, it recurs, and nobody's learning depends on it. If the tool is going to help you at all, it will show up there in the first week; if it does not help there, it is unlikely to help with anything harder.

A note on what this is not

None of this is an argument that teaching should be more efficient. Plenty of what looks inefficient in the job is the job: the conversation in the corridor, the second explanation, the time spent noticing.

The case for automating the paperwork is that it buys back attention for those. If it does not, and the time goes back into more paperwork, then the tool has made you a faster administrator, which nobody wanted.

That is also why we say plainly who pays for this site: nobody in this market does, which is the only reason to trust a page that tells you the number is half an hour rather than an afternoon.

The short version