What does being #1 in video editing rankings have to do with my office work?

This morning, after leg day, I slumped in my chair and saw AICPB's AI Product Rankings, specifically the video editing list. Veed.io topped it with millions of visits, followed by ClipChamp and InVideo. I stared at the list for a few seconds, and one thought popped into my head: this ranking tracks website traffic. High traffic only means good marketing or free traffic entry points; it has zero correlation with how usable the tool is.

This logic applies to office AI tools as well.

I started using WorkBuddy about two weeks ago, while also testing Doubao on company Feishu and Qianwen Office beta on DingTalk. I tried all three hands-on. Today I want to seriously discuss where WorkBuddy stands in office scenarios. Especially after seeing that video editing ranking, I'm more certain of one thing: You shouldn't choose office tools based on rankings; you should see if they actually get work done for you.

First, how I use WorkBuddy.

I run a fitness studio. When not teaching classes, I handle student profiles, training plans, and diet logs. Previously it was all manual. As students increased, the chaotic Excel formats drove me crazy. I started using Excel about a week ago but hadn't mastered it when I stumbled upon WorkBuddy. It has a feature to tidy messy table data into regular formats. That same day, I threw six months of training records into it, asking it to rearrange by date, student, muscle group, sets, weight, and notes.

That dataset was nasty. Some was typed manually, some copied from WeChat chats, some OCR'd from photos and skewed. Font sizes varied, and some rows had merged cells. WorkBuddy processed it, and about 80% was directly usable. The remaining 20% needed manual adjustment, mostly the skewed OCR parts where it sometimes read '10' as '1' or missed '0's. I accept this because manually organizing that batch used to take an afternoon; now it takes about twenty minutes.

Interestingly, I tried throwing the same dirty data to Doubao, asking it to organize it into a table. It could do it, and the resulting format was quite pretty, even slightly better-looking than WorkBuddy's default. So why did I ultimately stick with WorkBuddy as my main tool?

Because Doubao gave me a tidied table, while WorkBuddy gave me a reusable workflow.

This might sound abstract, so let's be specific. In WorkBuddy, you can save that table-tidying action as a workflow. Next time new data comes in, just select the same workflow and run it. Field mapping, format rules, and exception handling are hardcoded. With Doubao, you have to describe the requirements anew each time, and its understanding of context varies slightly every time. This time it puts a field in notes; next time it might put it elsewhere. In my testing, processing similar tables with WorkBuddy means no re-checking after the second time. With Doubao, I have to correct it roughly every third time.

This is what I said before: WorkBuddy is a dumbbell; Doubao and Qianwen are treadmills. You don't push a dumbbell, it doesn't move, but every push counts, and muscles stay grown. You jump on a treadmill and run, but when you stop, nothing remains. WorkBuddy's workflows are muscle memory; Doubao and Qianwen are treadmills you have to warm up on every day.

However, WorkBuddy isn't without flaws.

The first flaw is the entry barrier. When I first used it, translating the intent of "tidy data" into operations it understood took half a day. Unlike Doubao, where you throw things in, say "help me tidy this," and get a decent result immediately. WorkBuddy requires you to understand its logic first: data sources, processing rules, output formats—you need to know what you want for it to help you achieve it. This deters people completely unfamiliar with data, but for those with some basics, it's actually a good thing. It took me over a week to grasp the ropes, realizing the key is to break requirements down, letting it do one thing at a time rather than bundling everything together.

The second flaw is its insensitivity to "looks." The training plan tables I make for students export from WorkBuddy as plain, unadorned tables. Calling them "clean" is the highest praise. Doubao's templates are genuinely pretty, but flashiness doesn't help in office scenarios. Students need clear sets, weights, and reps, not comfortable color schemes.

The third flaw, which occurred to me while looking at that video editing ranking: Office AI tools nowadays love going "all-in-one," doing a bit of everything, but if you want to do serious work, you spend ages tweaking. WorkBuddy is relatively honest. It digs deep into documents, spreadsheets, emails, PPTs, and data analysis. Over these two weeks, the depth is sufficient, but the width is indeed narrow. If you want it to edit a video, it can't. But that's a video editing tool's job, not an office tool's.

I also tried using it to organize student diet logs these past two days. Those records were also a mess—memos, WeChat, Excel—all mixed. Some wrote "Breakfast: 2 eggs, 1 milk," others wrote "9:00 Egg2 Milk1." WorkBuddy can unify these into structured entries and aggregate approximate protein and carb values by day. But its protein calculation isn't accurate. It doesn't know egg sizes or whether the milk is whole or skimmed, and since I didn't provide that data, it didn't ask. I can't blame it; my data was incomplete. In the fitness industry, I track protein intake precisely in grams, so I manually reviewed that final table, taking half an hour. But doing it from scratch manually used to take two hours; it saved me an hour and a half.

My basic judgment now is this: WorkBuddy suits scenarios with fixed data structures where you're willing to teach it once and use it repeatedly long-term. I've built about six workflows for it: student profiles, training records, diet breakdowns, sign-in sheets, scheduling, and invoice summaries. Each workflow is configured once and then runs blindly thereafter. Who is it unsuitable for? People who use it once today and forget tomorrow, or people who don't even know what format they want. If you don't know what you want, it can't help you. Neither can Doubao.

Regarding that video editing ranking, honestly, I later flipped to the domestic growth rankings to see office AI placements, only to find the top spots still occupied by video, image, and chat products. Office tools rarely appear. This isn't surprising. Video editing tools are used once and left behind. You edit a clip today, maybe not tomorrow, so traffic is naturally high. Office tools stick to your desk. You use them daily, but you don't visit their official websites daily, so traffic doesn't spike. Therefore, ranking office tools by website traffic is a false proposition. Look at the rankings for fun, but don't take them seriously.

Back to WorkBuddy. This week I built a new workflow linking student photos/videos with training records, allowing me to quickly check what weight a student lifted last session before class. This workflow is still being tuned. Large video files slow down processing, so for now, I store videos separately and only keep filenames in the training records. Tools are like that; there are always places you can't reach, but where you can reach, it definitely saves time.

Tools will become more widespread, but people who can configure tools, set permissions, and manage operations are the valuable ones. I said this in my previous post, and I still feel the same way.