How to manage references across a multi-author paper without losing your mind

Five authors. One paper. Three reference managers between them, and at least one person who is "just going to keep the PDFs in a Drive folder for now." If you've ever been the corresponding author trying to compile the final bibliography the night before submission, you know the rest of the story. Duplicate entries. Inconsistent citation keys. A figure caption that cites Smith et al. 2019 — except one author has it as Smith 2019a and another has it as Smith2019_Nature.

This is the most googled workflow problem in academic writing for a reason. There is no clean answer. But there are honest tradeoffs, and some of them matter more than others depending on how your team actually works.

The three real options

Zotero

Zotero is, on balance, the best reference manager that exists. It's open source, free, and the browser connector is genuinely magical — one click on a publisher page and the metadata, the PDF, and the abstract all land in your library. Group libraries are free, and with Better BibTeX you get stable citation keys that survive collaborators renaming things. The annotation tools have caught up to the commercial options.

Where it breaks for multi-author work:

Storage is the friction. Free accounts get 300 MB. A real literature review burns through that in a week. Someone has to pay, or you fall back to "everyone hosts their own PDFs," which defeats the point.
The group library is a separate library. Authors have to remember to add things to the group, not their personal library. Half the time they don't, and you find out three weeks later when a citation is missing.
Sync conflicts are silent. Two authors edit the same item's notes on different machines, and one version quietly wins.
Citation keys still drift. Even with Better BibTeX, if one author is on Zotero 6 with one config and another is on Zotero 7 with another, the keys diverge.

If your team is small, technical, and willing to standardize on one config, Zotero is the right answer. Most teams are not all three.

Mendeley

Mendeley used to be the favorite. Then Elsevier bought it, killed the classic desktop client, and rebuilt it as "Reference Manager," which is slower and less capable. The social-network features are gone. Group libraries are still there but capped, and the export story is worse than Zotero's.

The honest take: if you're already on Mendeley and it works, fine. Nobody should be starting a new project on it in 2026. The tooling has not kept up, and the company that owns it has different incentives than the researchers using it.

The shared folder

Drive. Dropbox. OneDrive. A folder called papers/. Filenames like Smith_2019_attention_FINAL_v2.pdf.

Researchers know this looks bad. They do it anyway, because the activation energy is zero. No accounts to set up, no plugins to install, no group library to remember to add to. You drag a PDF in and your coauthor sees it.

The cost shows up later:

No structured metadata. Every citation has to be re-entered manually in whatever the final document uses.
No deduplication. The same paper gets uploaded by three people with three different filenames.
Annotations live inside individual PDF copies. They don't merge.
No way to ask "what have we cited so far?" without opening every file.

The shared folder is not stupid. It's a rational response to the overhead of the alternatives. But it pushes all the work to the end — to the corresponding author at submission time — instead of spreading it across the project.

What we built differently in Acteams

We didn't set out to replace Zotero. We set out to fix the part that Zotero, Mendeley, and the shared folder all leave broken: the literature isn't connected to the writing, the analysis, or the discussion.

In Acteams, every project has one library. Anyone on the project can add a paper — by DOI, by BibTeX, by RIS, or by dropping the PDF. There is no personal-vs-group split. If it's in the project, every author sees it, with the same metadata, the same citation key, and the same annotations.

A few things that fall out of this:

Annotations are shared by default. When one author highlights the methods section of a paper, the next author opens the same PDF and sees the highlight. Comments thread on top of it.
The AI assistant can read your library. "Which of our cited papers used a within-subjects design?" is a question you can ask, and the answer comes back with links to the specific papers and the passages it pulled from.
Citations attach to the document. When you write the manuscript in the same workspace, inserting a citation pulls from the project library — not from whatever your personal Zotero happens to have today. The bibliography is generated from one source.
Import is one-way and lossless. You can bring in your existing BibTeX file from Zotero or anywhere else. We don't ask you to give up your reference manager — we ask you to share one for the project.

When Acteams is not the answer

If you're a solo researcher with a Zotero library you've curated for ten years, you don't need us. Keep using Zotero. It's good.

If your collaboration is one shared paper with one coauthor and you both already use the same tool, you don't need us either. The overhead of switching costs more than the friction you're avoiding.

Where we think the math changes is at three or more authors, papers that span months, and projects where the literature, the data, and the draft all need to talk to each other. That's the case the existing tools handle worst, and that's the case we built for.

If that's your situation, try it on your next project. And if it's not — Zotero is still right there, and we mean that.