WorkBuddy for presentation PPTs: it doesn't make things pretty, it pulls you out of the pile of materials

Yesterday my boss was out, so as usual I spent a while scrolling Bilibili. It recommended a video about using AI to make PPTs, with a pretty grand title, saying that in 2026, using AI to make PPTs, a few universal prompts are the ceiling of efficiency. I clicked in and watched for a few minutes, then casually browsed a few related ones—manus, notebooklm, Kimi plus DeepSeek, and an open-source project PPT-Master with tens of thousands of stars. The pattern is about the same: give it a draft, and in a few minutes it spits out an editable PPT.

There was a line in the video I quite agree with. It said many people use AI to make PPTs, and the tools themselves aren't weak, but the output always has an AI flavor—flashy at first glance, but on closer look it's all template-y.

That's true. But my first reaction after watching was that this isn't my biggest headache. My headache is that I don't even have that draft.

1. The hard part isn't layout, it's before layout

Our department has to submit a quarterly work report this week. The boss's exact words were "make a PPT, present it at the meeting next Monday." Sounds simple. But when I opened the folder, the materials looked like this: a 3000-plus-word quarterly summary Word doc, accumulated from writing materials last month; an Excel ledger listing about twenty-some work progress items and numbers; four PDF notices and a scanned copy of a superior's document; and seven or eight screenshots saved from WeChat groups, some on-site photos, some data tables sent by others.

None of these materials can go directly into a PPT. The Word doc is too verbose, the Excel is all raw numbers, the PDFs are formal documents for the boss, and the screenshots are scattered everywhere. My old approach was to first spend two or three hours reading through all this, copying key points, writing an outline myself, and then making the PPT. Those two or three hours are the most annoying—pure grunt work, but you can't skip it.

This time I handed this part to WorkBuddy. I've been using it for about a month, previously for organizing submission packages and ledgers, but this was my first time using it for a report.

2. My steps, and the pitfalls I hit

From my testing, WorkBuddy did save me about two hours in this phase, but it's not one-click output—you have to use it in parts.

My approach was like this. First, create a folder and put all the original files in it, naming them as standardized as possible. This step is crucial. Last month I wrote a post about organizing red-header documents, and the lesson was: don't let it handle organizing, generating, and formatting all at once. The originals must be locked down and untouched.

Then let it do only one thing: read this folder, extract key points along four dimensions—work progress, data, existing problems, next steps—and output an outline. This step took about ten minutes, and the output was better than I expected. The numbers in the Excel mostly matched up, and several key statements in the PDFs were captured. A few items I had planned to fill in myself, it brought along.

Next was generating the PPT skeleton based on the outline. Here I hit a pitfall. The first time I directly asked it to produce a PPT from the outline, the layout was a mess—title hierarchy and body font sizes all wrong, clearly the template wasn't digested. Later I changed to having it first output a plain-text paginated structure, one title plus three to five points per page, and then I filled it into our unit's template myself. That was actually faster.

Finally, applying our unit's own master template—this step WorkBuddy can't help with, has to be adjusted manually. But the earlier skeleton already saved me most of the work. The whole process, from opening the folder to a draft ready for the meeting, took about two and a half hours. Previously this step took me four to five hours.

3. Where it stands compared to those AI PPT tools

Speaking of which, I need to be clear: WorkBuddy and the tools in those tutorials are not the same path.

Those tools solve the problem of "I have content, help me format it." The materials mentioned that manus's output can be directly edited and can specify template styles through preference settings, while notebooklm falls short on this point. That's their strength, I acknowledge. But their demos all assume you already have a clearly structured draft.

WorkBuddy's strength is in the earlier phase: it can first gather things scattered across Word, Excel, PDFs, and screenshots into usable content. My own feel is that its ability to handle mixed files is noticeably more stable—PDFs and tables thrown in together don't easily get messed up; step-by-step operation is more reliable than one-shot, doing one thing at a time yields much better quality; layout and aesthetics it's not good at, the template library is basically nonexistent, let alone animations; the generated text leans toward officialese, which for government materials is actually an advantage, but also easily becomes cookie-cutter.

So my judgment is: if the goal is a good-looking report PPT, I wouldn't use WorkBuddy—those dedicated AI PPT tools are faster; but if the goal is turning a pile of messy materials into content with a skeleton, WorkBuddy is the handiest I've used so far. Whether it's worth it depends on whether your work is stuck at that earlier step.

For people like me with scattered materials, mixed sources, and no desire to spend time on grunt work, it's worth it. For those doing external presentations, teaching courseware, or pursuing visual effects, it's not enough—you'd still need a dedicated tool set, or just do it manually. Also, there's a prerequisite: you have to be willing to spend ten minutes organizing the folder first. People who want one-click output will be disappointed with any tool.

This report goes to the meeting next Monday; we'll see the effect after submission. If the boss adds materials at the last minute, this split-process workflow will have to run again.