Papers AI

• Project-wide AI assistant that works with documents, datasets, references, code, and other project files
• Microsoft Word editing with DOCX support, comments, tracked changes, and rendered document styles
• Support for Markdown, Typst, LaTeX, Jupyter notebooks, CSV files, and Kanban boards
• Integrated Jupyter notebooks for running code and analyzing research data within projects
• Automatic citation management with reference libraries and continuously updated citations
• Integration with Zotero, Mendeley, and ReadCube for managing academic references
• AI-assisted writing that can answer questions, suggest edits, improve clarity, and work across project context
• Local AI model support through Ollama, LM Studio, vLLM, and OpenAI-compatible endpoints
• Real-time collaboration with comments, mentions, review history, and synchronized project editing
• Offline-first project storage with optional cloud synchronization and collaborative cloud workspaces

What Is Papers AI?

Papers AI is an AI-native research workspace developed by Digital Science and built by the team behind Overleaf. It brings academic writing, data analysis, references, code, notebooks, and collaboration into a single project environment.

Instead of limiting AI assistance to an individual document or chat window, Papers AI is designed to understand the wider context of a research project. It can work with drafts, datasets, notebooks, references, and other project files, helping researchers move between writing and analysis without repeatedly rebuilding context.

Project-Wide AI Context

A central feature of Papers AI is its ability to work across an entire research project. The AI assistant can access relevant project files and use their contents when answering questions or suggesting changes.

This project-level approach means researchers can ask questions about a document while also referring to associated datasets, references, notebooks, or other files. Users do not need to repeatedly explain the same project context to the assistant.

AI Writing Assistant

Papers AI includes an integrated AI writing assistant for research documents. Users can ask questions about their work, request edits, improve clarity, and develop sections while keeping the surrounding project context available to the assistant.

AI-generated changes remain under the user’s control. The platform allows users to accept, reject, or edit suggested changes rather than silently modifying research documents.

Microsoft Word Support

Papers AI supports real Microsoft Word DOCX documents. The workspace includes features such as comments, tracked changes, and rendered document styles, allowing researchers to work with Word documents inside the broader research environment.

This makes it possible to combine familiar Word-based academic writing with datasets, notebooks, references, and AI assistance in the same project.

LaTeX and Typst

Researchers who prefer technical or publication-oriented document formats can work with LaTeX and Typst alongside Word and Markdown. Papers AI is designed to accommodate different writing workflows rather than requiring every researcher to use a single document format.

The platform also supports compiling LaTeX and Typst projects locally in the browser, helping researchers work with technical documents without relying exclusively on external editing environments.

Markdown Writing

Markdown is another supported writing format in Papers AI. This provides a lightweight option for researchers who prefer plain-text-oriented document workflows.

Markdown files can exist alongside Word documents, LaTeX, Typst, notebooks, data files, and other project materials, allowing different parts of a research project to use appropriate formats.

Jupyter Notebook Integration

Papers AI integrates Jupyter notebooks directly into the research workspace. Researchers can run notebook cells while keeping their analysis close to the documents and other materials that depend on the results.

Outputs from data analysis can be reused in research documents as figures and tables, reducing the need to repeatedly export results and move them manually between separate applications.

Data Analysis

The workspace supports CSV data alongside notebooks and research documents. Researchers can keep datasets next to the code used to analyze them and the written material that explains the resulting findings.

This structure is intended to make the connection between data, analysis, and written conclusions easier to maintain throughout a research project.

Code and Writing Together

Papers AI combines code and research writing within the same project environment. Researchers can work with notebooks and datasets while simultaneously drafting papers, reports, or other research documents.

The integrated approach reduces the need to switch repeatedly between writing software, coding environments, reference managers, and AI assistants.

Reference Management

Papers AI includes reference management with automatically updating citations. Researchers can cite material directly from their research library while working on their documents.

The platform also connects with established reference-management services, allowing users to integrate existing academic collections into their Papers AI workflow.

Zotero, Mendeley, and ReadCube

Papers AI supports connections with Zotero, Mendeley, and ReadCube. These integrations allow researchers to work with references maintained through established academic reference-management platforms.

This can be useful for researchers who already maintain large citation libraries and want to bring those references into an AI-assisted research workspace.

AI Model Flexibility

Papers AI allows users to choose how AI is connected to their projects. In addition to the platform’s built-in AI services, users can connect their own models through OpenRouter or supported local AI applications.

Supported local model environments include Ollama, LM Studio, vLLM, and OpenAI-compatible endpoints. This provides researchers with additional control over which AI system processes their project material.

Local AI Models

Users who prefer to run AI locally can connect compatible local models rather than relying exclusively on cloud-based AI. This option is particularly relevant to research workflows where users want greater control over where project information is processed.

Model compatibility depends on the connected provider and configuration, so researchers should check current platform requirements before selecting a local model.

Offline-First Workspace

Papers AI is designed as an offline-first research workspace. The platform states that projects can remain on the user’s device unless cloud synchronization or collaboration is enabled.

LaTeX, Typst, MyST, and Jupyter workflows can run locally in the browser, allowing researchers to continue working without depending entirely on a persistent internet connection.

Local Project Storage

Projects are stored locally by default rather than automatically being synchronized to the company’s servers. Users can choose to save projects to the cloud when they need access across devices or want to collaborate with others.

This local-first model gives researchers more control over where their working files are stored, although users should review the platform’s current data and security policies for their specific research requirements.

Cloud Collaboration

Papers AI also supports cloud-based collaboration for projects that need to be shared. Multiple researchers can work on the same project while seeing changes in real time.

Cloud collaboration allows teams to combine writing, analysis, references, and project discussions without maintaining separate copies of the research workspace.

Real-Time Editing

The collaboration environment supports simultaneous editing and presence indicators. Papers AI states that multi-user editing can operate across formats including DOCX, LaTeX, and Jupyter notebooks.

This allows coauthors and research teams to work together without relying solely on manually exchanged document versions.

Comments and Review

Papers AI includes threaded comments, mentions, and review workflows. Collaborators can discuss specific parts of a project, respond to feedback, and resolve review threads.

The platform also maintains reviewable history so that edits can be attributed and reversed when necessary.

Research Project Organization

A Papers AI project can contain multiple types of research material. Documents, datasets, notebooks, reference files, and other project resources can be kept together instead of being distributed across unrelated applications.

This project-based structure is intended to preserve relationships between research materials and give the AI assistant a broader context when helping with the work.

Export Options

Papers AI supports exporting research documents in formats including PDF, DOCX, and Markdown. This allows users to move completed or intermediate work into formats suitable for sharing, reviewing, or further editing.

Export capabilities complement the platform’s support for multiple native research formats and help researchers maintain flexible publishing workflows.

Academic Research Workflow

A typical workflow can begin with creating a research project and adding documents, references, datasets, and notebooks. Researchers can then write and edit documents while using AI to analyze project materials, answer questions, or suggest changes.

Data can be analyzed in Jupyter, results can be incorporated into documents, citations can be managed through connected reference libraries, and collaborators can review the project through the same workspace.

Privacy and Data Control

Papers AI emphasizes local-first storage and states that projects remain on the user’s device by default. Cloud synchronization is optional for projects where users need cross-device access or collaboration.

The platform also states that user projects are not used to train AI models. Its terms nevertheless note that AI-generated outputs may contain incorrect information and should be evaluated by users for accuracy and suitability.

Pricing and Plans

Papers AI offers Free and Plus plans. The Free plan costs £0 and includes a generous AI allowance, local and cloud work, local AI model connections, offline operation, and collaboration.

The Plus plan costs £10 per month when billed monthly. It includes everything in Free along with five times more AI usage, longer AI conversations, and in-depth analysis.

Both plans include unlimited projects, Word, Markdown, Typst, and LaTeX writing, Jupyter analysis, reference management, connections with Zotero, Mendeley, and ReadCube, and export to PDF, DOCX, and Markdown. Usage limits apply.

Who Is Papers AI For?

Papers AI is designed primarily for researchers, students, academics, scientists, and research teams who need to combine writing, data analysis, coding, references, and collaboration.

It can be particularly useful for projects where written research depends on datasets, computational analysis, citations, and multiple supporting files.

Built by Overleaf’s Team

Papers AI was built by the team behind Overleaf and is operated by Digital Science. The product extends the research-writing expertise associated with Overleaf into a broader workspace covering writing, data, code, references, and AI assistance.

The platform is separate from the former Papers reference-management product associated with ReadCube. Papers AI is positioned as a new research workspace rather than a continuation of that product.

Frequently Asked Questions

What is Papers AI?

Papers AI is an AI-native research workspace that combines writing, data analysis, code, references, project files, and collaboration in one environment.

Is Papers AI free?

Yes. Papers AI has a Free plan with core workspace, AI, writing, reference, collaboration, and offline features. A paid Plus plan provides additional AI usage and analysis capabilities.

How much does Papers AI cost?

The Plus plan currently costs £10 per month when billed monthly. The Free plan costs £0.

What file formats does Papers AI support?

Papers AI supports Word DOCX, Markdown, Typst, LaTeX, Jupyter notebooks, CSV files, and other project materials.

Can Papers AI run code?

Yes. Researchers can run Jupyter notebook cells within their projects and use analysis outputs as figures and tables in their research documents.

Does Papers AI support citations?

Yes. Papers AI provides reference management with automatically updating citations and connects with Zotero, Mendeley, and ReadCube.

Can I use my own AI model with Papers AI?

Yes. Users can connect an OpenRouter key or local AI models through supported environments such as Ollama, LM Studio, vLLM, and OpenAI-compatible endpoints.

Does Papers AI work offline?

Yes. Papers AI is designed as an offline-first workspace, with projects and several research workflows capable of running locally on the user’s device.

Does Papers AI support collaboration?

Yes. Users can collaborate in real time, add comments and mentions, review changes, and maintain project history when using cloud collaboration.

Is Papers AI the same as Papers by ReadCube?

No. Papers AI is a research workspace operated by Digital Science, while the former Papers reference-management product is now known as ReadCube.

Who is Papers AI designed for?

Papers AI is designed for researchers, students, academics, scientists, and research teams working with academic documents, data, code, references, and collaborative projects.

Share Your Experience: