One research workspace for all your writing

Your research, your writing, your data, all in one place. With AI that has your whole project in context.

Get Started

calibrated-disagreement-2026

draft.docx

results.tex

3 Methodology

3.1 Research design

A mixed-methods approach was adopted, combining structured surveys with semi-structured interviews across three independent cohorts.

3.2 Survey instrument & sample

We administered the calibrated questionnaire to n = 412 participants across three cohorts, following the protocol described in [Park & Liao, 2024].

Responses were stratified by self-reported expertise tier (A1–A4) and weighted against the prior distribution from [Anand et al., 2023]. Internal consistency was assessed via Cronbach's α.

Open responses were coded by two independent annotators; inter-rater reliability reached κ = 0.81.

AI ASSISTANT

AI panel

Ask questions, request edits, or get help with your document.

000Synced

Ln 12412 words2 min readWordComments 3Private

Scroll

Formats

One workspace. Multiple formats.

Write in Word, Markdown, LaTeX, and more. Manage datasets, and run code in Jupyter notebooks.

3 Methodology

3.1 Research design

A mixed-methods approach was adopted, combining structured surveys with semi-structured interviews across three independent cohorts.

3.2 Survey instrument

We administered a 32-item Likert questionnaire to n = 412 participants stratified across four expertise tiers (A1–A4). The instrument was adapted from the calibrated battery of Park & Liao (2024) and validated against an independent pilot cohort (n = 48).

Internal consistency was assessed via Cronbach's α; per-tier reliability ranged from α = 0.78 to α = 0.91, indicating acceptable scale coherence.

α = N · c̄ / (v̄ + (N − 1) · c̄)

000Synced

Ln 12412 words2 min readWordTrack Changes · ONComments 3Private

Plan

Track your project, not just your paper.

Kanban boards and Gantt timelines live inside your project, right next to your drafts, data, and notebooks.

▽Filter cards...◇ Labels ▾9 cards

Backlog2

Define inclusion criteria for pilot dataset

methodsP1

Request updated ethics approval letter

adminP2

In Progress2

Clean survey responses from cohort B

dataP1

Draft methods section for preregistration

writingP1

Review2

Review literature table for duplicate studies

literaturereview

Validate figure captions against journal guidelines

figuresreview

Done2

Import Zotero library into project references

references

Create shared analysis notebook skeleton

notebookanalysis

Assistant

AI that knows your whole project.

Add your notebooks, drafts, and references once. No re-uploading, no re-explaining, no copy-pasting.

AI PANEL · draft.docx

Skim the project. What are the open questions in §3?

✓ Read draft.docx · 38 pages

✓ Indexed 14 sections, 23 references

§3.2 has a TODO on tier breakdown, and §3.4 cites a paper that isn't in your refs.bib.

›Reading your project…

Collaborate

Write anywhere. Online or off.

Edit alongside coauthors, comment in margin, and see exactly who changed what. Built on a conflict-free protocol that syncs when you reconnect, so collaboration works on a plane, too.

Live cursors

Follow along in real-time

Multi-user editing with sub-50ms presence in any format: .tex, .docx, .ipynb.

Margin comments

Threaded review

Resolve, reopen, @-mention. Comments survive format conversion.

Tracked changes

Reviewable history

Every edit attributed and reversible, including the assistant's.

draft.docx

3 ONLINE

● synced3 collaboratorsLast edit · Aisha · 1s ago

Private by design

Your code belongs to you.

Papers AI is an offline-first, private workspace. Files live on your device, compilers run in your browser, and collaboration only happens when you ask for it.

Your files stay on your device

Projects live in your browser's local storage or a folder you pick. Never on our servers unless you share.

Compiles in your browser

LaTeX, Typst, MyST, and Jupyter all run on your machine. No server queue, no round trip. Works offline too.

Bring your own AI model

Use our AI or point AI at Ollama, LM Studio, or any OpenAI-compatible endpoint. You choose what reads your work.

Sync only when you invite

Real-time collaboration is opt-in and project-scoped. Invite people, and only those projects sync.

Connected

CLOUD SYNC

In [3]:

from sklearn.metrics import cohen_kappa_score
cohen_kappa_score(rater_a, rater_b)|

Out:

0.81
▸ ran on cloud kernel · 22 ms

In [4]:

# model: ai
ai.ask("Summarise §3.2")
|

Using: AI cloud assistant

In the wild

What researchers are saying.

"I used to keep my analysis in Jupyter, my figures in a folder, and my paper in Word. Now the notebook runs against my dataset right next to the draft — my results and my prose stop drifting apart."

Lena Okafor(Postdoctoral Researcher, Computational Neuroscience)

"When a reviewer asks me to re-run an analysis, I update the cell and the figure regenerates in the same place the manuscript lives. No more re-exporting plots and pasting them back in."

Marco Bianchi(PhD Candidate, Experimental Psychology)

"The assistant reads my survey data and my refs, not just the paragraph I paste in. It pulled the tier counts straight from my CSV into the methods section — the part that usually eats my whole afternoon."

Priya Raghavan(Research Scientist, Epidemiology)

"Half my coauthors live in Word and I live in LaTeX. This is the first tool where we edit the same document without my tracked changes turning into chaos."

Daniel Hofmann(Associate Professor, Applied Mathematics)

"My interview transcripts, my coding, and my write-up finally sit in one project. I stopped emailing myself versions just to move between apps."

Sofía Castillo(Qualitative Researcher, Sociology)

"I draft on research cruises with no signal and it syncs the moment I am back online. Nothing I wrote at sea has ever been lost."

Yuki Tanaka(Graduate Researcher, Marine Biology)

Get Papers AI

Move your next paper into Papers AI.

Bring a project or start from a template. Your new workflow is one import away.