Automating the file chores you do every week, without writing code

The tasks worth automating are not the interesting ones. They are the small chores repeated weekly: resize these for the site, rename these by date, convert these to PDF, move last month's into a folder.

Both major systems ship a visual automation tool that does all of it, and both are largely unused because they look like programming and are filed under settings.

The four chores worth doing first

Resize and convert a folder of images. Pick a folder, output at a fixed width, convert to a modern format, save to a subfolder. Three steps in either tool.

Rename by date and project. Take a selection, prefix an ISO date, add a project code, lowercase the rest. This is the single highest-value automation for anyone whose downloads folder is an archaeological record.

Convert documents to PDF. A folder of images or documents into one PDF, in name order — which is why a consistent naming scheme comes first.

File by age. Move anything older than 90 days from Desktop or Downloads into a dated archive folder. Run monthly.

None of these is impressive. Together they are most of what people do by hand.

macOS — Shortcuts

Shortcuts replaced Automator and now runs the same automation on Mac, iPhone and iPad.

• Build: open Shortcuts, add actions from the sidebar, drag to order. Each action takes the previous one's output.
• Run on files: enable Use as Quick Action so it appears in Finder's right-click menu for a selection.
• Run on a schedule or a folder: Shortcuts handles time and event triggers; Folder Actions still exist for watch-a-folder behaviour.
• Useful actions to know by name: Get Images from Input, Resize Image, Convert Image, Rename File, Format Date, Make PDF, Move File.

Windows — Power Automate

Power Automate's desktop flows do the same locally, with a recorder for the fiddly parts.

• Build: drag actions into the flow, or record yourself doing it once and then edit the steps.
• Run on files: flows accept input parameters, and can be triggered from a desktop shortcut.
• Run on a schedule: built in, or through Task Scheduler.
• File actions cover copy, move, rename, and iterating over a folder; image work is available through actions or a called command-line tool.

The rules that keep automation from causing damage

1. Work on a copy the first time. Every time. An automation that renames wrongly renames everything wrongly, at speed, with no undo.
2. Output to a new folder rather than overwriting the input. Disk is cheap and mistakes are not.
3. Run it manually a few times before attaching it to a folder or a schedule. A broken automation you trigger is a nuisance; a broken one that fires on its own is a mess you find later.
4. Handle the empty case. A flow that runs on an empty folder should do nothing, not error or produce an empty artefact.
5. Name the automation for what it does, not what it is for. In a year, Resize to 1200px and convert to WebP is legible; Blog images is not.

Automate the thing you have already done by hand more than five times. Automating in advance is how people spend an afternoon building something they use twice.

When to reach for something else

• Once only — do it by hand. Building the flow costs more than the task.
• Genuinely large batches, thousands of files, where a command-line tool is faster to write and faster to run.
• Anything needing a decision per file that you cannot express as a rule. If the rule requires judgement, you are the rule.
• Cross-machine or team-wide, where a script in a shared repository beats an automation living on one laptop.

The point of these tools is not to replace scripting. It is that most household file chores never justified a script, so they never got automated at all.

What is file automation?

File automation is a saved sequence of steps that a computer applies to files on your instruction — rename these, resize those, convert this folder, move anything older than a date. It is not scripting, though it can call a script, and on both major systems it is built by dragging actions into order rather than by writing code.

The distinction from a utility app is ownership: file automation you built does exactly what you specified, runs offline, needs no account, and cannot be discontinued.

Which tool for which situation?

The middle row is where most people stop too early: file automation on a laptop is excellent for the 20-file chore and the wrong shape for the 5,000-file job.

How do you avoid destroying a folder?

By treating the first run as an experiment, every time. The 5 rules that matter:

1. Work on a copy the first time. There is no undo, and a wrong rule is applied at speed.
2. Output to a new folder rather than overwriting the input.
3. Run it manually several times before attaching it to a folder watcher or a schedule.
4. Handle the empty case, so a run over an empty folder does nothing rather than erroring.
5. Name it for what it does, not what it is for.

Start with the chore you have already done by hand more than 5 times. Building file automation in advance is how an afternoon disappears into something used twice.

See also our guides to file tools, text tools worth automating, and the free utilities that cover the rest.

Where to read the documentation

Both tools are better documented than their reputation suggests. The Shortcuts user guide covers actions and triggers on Apple platforms, and Microsoft's Power Automate documentation covers desktop flows on Windows, including the recorder.

Neither is long. An hour with either is enough to build the 4 chores above, and file automation that saves 10 minutes a week repays that hour inside a quarter — which is the honest reason to bother, rather than any argument about efficiency in the abstract.

The chores worth automating are boring by definition. That is what makes them safe to hand over: a rule you can state completely is a rule a computer can apply, and everything else stays yours. Everything that needs judgement stays with you, which is the correct division of labour and the reason this scales. Start with one chore this week and the second will build itself in ten minutes. The second automation always takes a fraction of the time the first one did, which is the real argument for starting.