Physix FrontierAfter using WorkBuddy for a week, I uninstalled Zotero and EndNote—reference managers may be heading in the wrong direction
Honestly, I didn't expect it to turn out this way either.
I read that CSDN comparison post last week comparing 8 tools, from Zotero to EndNote to Scholard. It covered quite a lot. But after reading it, I had one feeling: These people are still competing on whose literature management is better, but no one asks a fundamental question—what are you managing literature for?
To write papers. Not to organize a library.
When I was a first-year master's student, I also tinkered with Zotero plugins. I spent three days configuring it, only to find that copy-pasting Chinese PDFs resulted in nothing but spaces. EndNote was even harder; my advisor insisted I use it, and I installed it four times before it finally ran. Later I used Little Green Whale (Xiao Lü Jing); the translation quality was okay, but every time I wanted to extract a paragraph from a paper into my own draft, I had to manually copy, paste, and reformat, constantly switching windows.
What I really needed wasn't "managing literature," but "turning literature into part of the paper."
First, a Counter-Intuitive Conclusion: Literature Management Tools Might Not Need to Exist
Don't rush to criticize me.
Look at those 8 tools; their functional structures are similar: literature import, PDF reading, translation, citation formatting. But their core logic is "storing and finding," not "using and writing."
Zotero stores well, but when I finish reading a paper and want to extract its methods, data, and conclusions to put into my own paper—Zotero can't do that. EndNote has comprehensive citation formats, but if I want AI to help summarize the commonalities across 10 papers—EndNote can't do that either.
A friend said something I found quite accurate: Literature management tools manage literature, not knowledge.
And WorkBuddy's entry point happens to be exactly here.
What WorkBuddy Actually Does—My Understanding
Officially, it's described as an AI office efficiency tool capable of handling documents, spreadsheets, emails, PPTs, data analysis, and file organization. Sounds miscellaneous, but after using it, I found its core capabilities are actually just three:
1. Document Understanding: Not just OCR recognition, but understanding document structure and logical content, then extracting, summarizing, and comparing.
2. Task Decomposition: It can break down big tasks like "writing a paper" into a workflow of "read literature → extract viewpoints → write review → list references," providing tools for each step.
3. Multi-format Output: Input is PDF, output can be Word documents, Excel tables, PPT outlines, or even emails.
I compared the key differences between WorkBuddy and traditional literature management tools:
| Dimension | Zotero / EndNote | WorkBuddy |
|---|---|---|
| Core Function | Literature ingestion, citation format management | Document understanding, knowledge extraction, paper generation |
| Chinese Literature Handling | Poor native experience, requires plugins | Built-in Chinese optimization, PDF re-layout + paragraph-by-paragraph translation |
| AI Assistance | Requires third-party plugins, cumbersome configuration | Native integration, auto-extracts objectives/methods/conclusions |
| Output Capability | Citation formats only | Can generate paper paragraphs, reviews, PPTs, emails |
| Learning Cost | High (Zotero plugins, EndNote interface) | Low (download and use, interface similar to office software) |
| Suitable For | Heavy library builders | Research newcomers, paper writers, interdisciplinary researchers |
My judgment: Zotero suits "collector" type researchers, while WorkBuddy suits "producer" type researchers.
My Complete Workflow Using WorkBuddy to Write Papers (With Data)
Background: I'm a third-year master's student in humanities. My thesis topic is on "digital labor," and I frequently read both English and Chinese literature. Previously, using Zotero + Little Green Whale, every time I wrote a paper I had to:
1. Find literature in Zotero
2. Open the PDF
3. Go back to Word to write
4. Return to Zotero to copy citations
Switching back and forth wasted at least 20 minutes per paper.
After using WorkBuddy, I changed my process to this:
Step 1: Throw the literature into WorkBuddy
I dumped 15 PDFs in at once. It automatically identified the author, title, abstract, and keywords for each. I spent about 3 minutes checking, with an accuracy rate of roughly 95%. The only two errors were two English papers with subtitles, where it treated the subtitle as body text.
Step 2: Have WorkBuddy Extract Core Content
I clicked "Extract Key Information," and it automatically generated a table for each paper containing: research question, method, sample, main findings, and innovations.
My tests showed extraction time per paper was about 10 seconds. For 15 papers, that's 150 seconds, less than 3 minutes total.
Step 3: Have WorkBuddy Help Write the Literature Review
I gave a prompt: "Summarize the current state of research on 'control mechanisms in digital labor platforms' across these 15 papers, organized by 'control means → effects → controversies'."
WorkBuddy generated a review of about 800 words. I spent 30 minutes revising, mainly adjusting the tone, adding viewpoints from 2 papers I hadn't uploaded, and correcting 1 citation error.
Comparison with before: Writing an 800-word literature review used to take at least 3 hours—1 hour reading, 1 hour drafting, 1 hour formatting.
On this task, I saved about 2 hours.
But I Also Have to Be Honest: WorkBuddy Isn't a Panacea
I used it for about a week and encountered several issues:
1. Journal citation formats sometimes have errors
In the APA format it generated, the DOI links for 2 papers were wrong. I checked, and it might be because the metadata it cited came from different databases, causing a field mismatch. Manual correction is needed.
2. Deep understanding of Chinese literature is worse than English
For the same 15 papers, extraction accuracy for English literature was about 95%, but dropped to around 80% for Chinese literature. Mainly, it struggled to understand certain expressions in the "research methods" section of Chinese papers. For example, it recognized "Grounded Theory" as "Basic Theory."
3. I admit I'm being a bit hindsight-biased
WorkBuddy's "literature review" function essentially helps you logically reorganize extracted key information. It won't help you with "critical thinking" or judge whether "a paper's conclusion is flawed." You still have to do that yourself.
So my judgment is: WorkBuddy is suitable for "acceleration," not for "replacement."
About That CSDN Comparison Post—Why I Feel It Missed the Point
That article listed all 8 tools, covering everything from ease of use to translation quality to AI assistance. But the problem is, it treated "AI assistance" as an independent feature point, rather than a core capability.
Zotero's AI assistance is "requires third-party plugins," while ScholarRead's is "native integration." But regardless of which, the logic is "help me read literature"—not "help me write papers."
I suspect the future direction of academic tools isn't "literature management," but "knowledge production." You don't need a tool to store 1,000 papers; you need a tool to help you find the 3 sentences from those 1,000 papers that you need for your writing.
What WorkBuddy is currently doing aligns perfectly with this direction. It connects the two stages of "reading literature" and "writing papers" using AI.
Finally, An Open Question
I'm actually unsure if I am WorkBuddy's target user.
If I were a first-year master's student, I might continue using Zotero, because at that stage I only needed to "store literature," not "write papers." But if I'm a third-year master's student, PhD candidate, or young faculty member, I think WorkBuddy is much more useful than Zotero.
So here's the question: What should the future of academic tools look like? Should we continue optimizing the ability to "manage literature," or shift towards helping people "write papers"?
Currently, I'm betting on WorkBuddy. But I'm not sure if I'll still feel this way in a year.
What do you guys think?