For research groups
The first pass, done before the draft reaches your desk.
If you coordinate a group, you already know what you will find in the draft before you open it: the citations that are not in the reference list, the method that never says on what criterion it stopped, the discussion that claims more than the data shows. It is mechanical work and it comes round again with every new person. Can I Publish does it first, so your reading starts where you actually add something.
The group works on the same diagnosis
A report that arrives by email is read once and then lost. Here the diagnosis is where the group works: it gets shared out, marked as resolved and compared against the previous version.
- Work is shared out by finding, not by manuscript
- Each finding is assigned to the person who is going to fix it, and that person sees it in their own task list. They are notified by email, with notices batched so that an afternoon of handing out work does not turn into twenty emails.
- The assignment survives into the next version
- When a new version is uploaded, every finding is compared against the previous one — resolved, still there, new or a regression — and anything still open keeps whoever it was assigned to. Nobody hands out the same work twice.
- Real co-authorship, not a shared account
- You invite by email, each person signs in with their own account and role, and is added as a co-author to the projects that concern them. You see the group's manuscripts; each member sees what they take part in.
Group memory: your criteria, written down once
Every group has its own particulars, and they almost always live in the head of whoever coordinates it. Here they are written into a document — what conventions you follow, what you demand of a method section, where you usually publish, what you tend to forget — and they go into every diagnosis in the group, for everyone, in every version.
## Conventions
- We cite in APA 7. In this group we write «participants», never «subjects».
- We work with small samples by design; if the sampling is justified, it is not a limitation.
- We almost always submit to the same two journals in the field.
## Where we tend to slip
- We tend to forget to state the saturation criterion — always check for it.
- We claim causality in the discussion with correlational designs.
Illustrative examples
- It stops flagging what you already decided
- Your conventions go into the analysis marked as settled decisions, not as candidate problems. A diagnosis that argues with every style choice the group has made is a diagnosis nobody opens twice.
- And it looks hard at what you already know gets past you
- The same document has a section for the opposite: «where we tend to slip». There you write the lapses that keep coming back — the saturation criterion that never makes it into Method, the causality that creeps into the discussion — and the diagnosis checks them first in every manuscript from the group. It is not a checklist that ticks itself: if the manuscript handles one well, it does not appear, because every finding still has to bring the literal sentence from the text that supports it.
- Journal fit takes your habits into account — and contradicts them when it should
- If the memory says where you publish, that journal is always addressed explicitly, whether or not it comes out among the best fits. And if it is a bad idea for that manuscript, it says so in plain words: habit does not replace real fit, but you should be told, not left to work it out from a table.
- Whoever joins inherits the criteria without being told twice
- What a PhD student learns today through rounds of your corrections is written down and applied from their first manuscript. Any member of the group can edit it: it is a living document, not an admin setting.
Calibrated to how your group does research
The study design is detected first — quantitative, qualitative or mixed — and the matching rubric is applied. This is not an implementation detail: a qualitative manuscript is not judged by the standards of a clinical trial, and saturation, reflexivity or transferability do not show up in a system built for something else.
The rubrics are built on 292 open peer review reports on 118 published papers, coded one by one. We accept manuscripts in English and Spanish; reports are currently delivered in Spanish.
The full method: the corpus, inter-rater reliability and how the benchmark was measured.
Can I Publish vs other LLMs
What happens to your people's manuscripts
It is the first question any coordinator asks, and the answer is the same one we give everywhere:
- The original file is deleted once the text has been extracted.
- The extracted text is deleted after 30 days.
- Nobody trains models on your manuscripts, with contractual zero retention.
For the paperwork with your university, a data processing agreement (DPA) is available. The full privacy policy sets out what is kept, for how long and on what legal basis.
What it does not do, said before you ask
- It does not replace your judgement: it takes the mechanical pass off you so that your reading is the one that counts.
- It does not guarantee acceptance. Nobody can, and anyone promising it is selling something else.
- It does not fix style or grammar. There are better and cheaper tools for that.
- Our evidence comes from open, published review; about double-blind review we claim nothing.
How you get in
We are opening up group by group, not one person at a time: we want to see the product working inside a real team, with its versions and its co-authors. Leave us your email, tell us you coordinate a group, and we will talk about how it fits your workflow.