Cleaning Up Canva AI's File Mess with WorkBuddy: I Got the Workflow Running Smoothly
Community Discussion · General

Cleaning Up Canva AI's File Mess with WorkBuddy: I Got the Workflow Running Smoothly

Pixel DustPixel Dust1d ago2026/11/17 86 views

At the AI image generation step, you don't really need a designer anymore. The trouble is that after generating, dozens of filenames lie in the download folder like garbled text.

Last night I came across a Cantonese Canva AI tutorial about using it to make IG posts, posters, and decks. I watched it. The tool itself is fine—type a topic into Magic Write and it produces an outline and layout, good enough for non-design roles. I've used Canva for two months, and this kind of work is indeed easy. But I zoned out halfway through: someone in the comments asked how to manage files after exporting, and nobody answered. I know that problem all too well.

Last month I took a small job—the client wanted a full autumn/winter asset set: one key visual poster, six social media square images, and a twelve-page proposal deck. Canva AI generated the images in about forty minutes. Organizing those files took me an evening. Later I reran that part with WorkBuddy, and now I can get it under half an hour.

Build the skeleton first, then talk about automation

My setup is macOS, WorkBuddy desktop version, updated a few times over three months—can't remember the version number. This won't necessarily apply to everyone.

The first thing is to manually create the folders in Finder, then open WorkBuddy. 01_Key Visual, 02_Social, 03_Deck, 04_Source Files, 05_Delivery. This step can't be skipped. I once wrote a post complaining about this—the tool will, by its own understanding, create a bunch of folders with the same name for you, a whole screen of duplicates, looking like a misaligned grid.

Once the skeleton is built, open WorkBuddy, click Tasks in the left sidebar, New Task in the top right, and choose File Organization. Here it asks you to specify a working directory—choose that root folder. It has "Auto-create subfolders" checked by default; uncheck it.

Then write the rules—this is the most critical step.

In the rule input box, write one rule per sentence; don't use "and" or "also." This time I wrote it like this: files whose names contain post move to 02_Social; those containing deck or brief move to 03_Deck; those with extension .psd or .ai move to 04_Source Files; the rest stay in the root directory. Four rules; when done, click Preview in the bottom right.

The Preview button is important. It doesn't actually move files, just shows you the move list. The first time I used it I didn't know it existed and clicked Execute directly, and three posters ended up in the source folder. Now my habit is to preview twice and confirm there's nothing under 01_Key Visual in the list before executing.

SubAgent is queuing, not running in parallel

After organizing files, the next step is generating the delivery list. I use SubAgent for this—started about two weeks ago, still figuring it out.

SubAgent is a temp worker dispatched by the main Agent; the main Agent doesn't do the work itself, it just splits tasks, dispatches work, and collects results. It dispatches in the SubAgent panel on the right, one at a time, then dispatches the next when done. Don't think about opening three at once to run in parallel—I tried, and with two tasks running simultaneously they start bleeding into each other, with fields from the previous task mixed into the output.

The task brief has to be written so it can be verified. Something like "generate a table" won't do; you have to write "generate an xlsx with four columns: filename, pixel dimensions, purpose, status; status defaults to To Be Confirmed; do not merge cells." The more rigidly you write it, the less rework.

When delivering to the client, I set the 05_Delivery folder to shared with View only permission. The client can download but not modify. Revisions go through me; after revising I overwrite the file with the same name, and the link stays the same. This is easier than sending a zip each time, and you don't have to worry about the client scribbling annotations all over the files.

The remaining pitfalls

WorkBuddy's undo isn't well done. Once a file is moved wrong, the undo button sometimes doesn't work, and you have to drag it back from the target folder yourself. So I never dare skip the preview step.

The interface really isn't pretty either. That line spacing in the task list—I want to adjust it every time I look. But I've accepted it; as long as it works.

AI handles generating images, WorkBuddy handles putting them back where they belong—don't expect one tool to do both.

2 replies

?
Ctrl + Enter to reply
I am Dapeng

Listen to me, this requirement is simple—that rule "everything else stays in the root directory," if Canva exports filenames with spaces that don't match, don't we have to drag them manually again?

Engineer Xue

I believe the SubAgent cross-contamination thing, same problem as Copilot when running multiple tasks in parallel—if the task brief fields aren't hardcoded, they contaminate each other.