Notebook Collaboration Platform

Part of the worlds-biggest-software-project initiative.

An AI-native, open-source collaborative notebook platform that is Jupyter-compatible and combines real-time co-editing, version control, and scheduled execution.

The Notebook Collaboration Platform is a cloud-hostable and self-hostable workspace for data science and ML teams who need more than a vanilla JupyterHub deployment but do not want to be locked into a proprietary SaaS. It targets the gap between bare-metal Jupyter (powerful but DevOps-heavy) and polished commercial tools (Deepnote, Hex, Databricks) by combining a .ipynb-compatible multiplayer notebook with AI capabilities that understand the full notebook state.

Why Notebook Collaboration Platform?

JupyterHub gives full control but ships no built-in collaboration UX, no scheduling, and no AI assistance — teams pay for that polish in DevOps time.
Deepnote, Hex, and Databricks deliver strong collaboration but are proprietary SaaS with no self-hosting path; pricing in the contested mid-market sits at roughly $24–$40/editor/month.
Google Colab is free but throttles compute, has no team management, and offers Python only.
Briefer is the only fully open-source tool with native multiplayer plus dashboards, but its community, integrations, and AI features are still early-stage.
Across every incumbent, raw cell diffs, manual scheduling, and isolated single-notebook collaboration leave clear room for an AI-native open alternative.

Key Features

Real-Time Collaborative Notebook

Multiplayer editing with presence indicators using Yjs/CRDT (BSD-compatible)
.ipynb import and export for portability across the Jupyter ecosystem
Mixed Python and SQL canvas in a single notebook
Cloud-hosted persistent storage with share-by-link
Role-based access control (viewer, editor, admin)

Versioning and Git Integration

Automatic version history with per-change audit trail
Git integration with commit, branch, and visual diff viewer
Semantic version diffing: plain-language summaries of what changed between runs
Compatibility with existing Git-based review workflows

Scheduling and Orchestration

Scheduled notebook execution with email or webhook notifications
REST API and webhooks for triggering runs from external systems
Foundation for future cross-notebook pipeline orchestration

AI-Native Assistance

Code completion that reads full notebook state — variables, schemas, prior outputs
Automated narrative generation: prose suggestions between code blocks
Semantic diff summaries describing what changed and why it matters
Output anomaly detection alerting collaborators when results deviate from expected ranges
Intelligent scheduling that predicts runtime and right-sizes compute before the job starts

Publishing and Extensibility

Notebook-to-dashboard publishing for non-technical stakeholders
Planned R and Julia kernel support beyond the Python and SQL MVP
Self-hostable open-source edition for air-gapped or on-premise deployments

AI-Native Advantage

Unlike incumbents that bolt generic autocomplete onto a notebook, this platform treats the notebook's live state — variable values, dataframe schemas, and prior cell outputs — as first-class context for the model. That same context unlocks features no current tool offers: plain-language semantic diffs between versions, narrative-gap detection that suggests explanatory prose between code blocks, anomaly detection on cell outputs, and AI-aware scheduling that predicts runtime from historical execution data before provisioning compute.

Tech Stack & Deployment

The platform is designed for both managed cloud and self-hosted deployment, with parity between the two so teams can migrate without rewriting workflows. The notebook layer builds on the Jupyter ecosystem: nbformat for the document model, .ipynb for portability, and Yjs CRDTs for real-time collaboration (the same approach used by JupyterLab's jupyter-collaboration extension). Authentication uses OIDC / OAuth 2.0 for enterprise SSO. Reproducible environments follow the Binder / repo2docker pattern via Docker images. Integration surface includes a REST API and webhooks for external orchestration.

Market Context

The broader data science platform market is valued in the tens of billions, with the notebook-specific segment estimated at several hundred million dollars and growing as data teams scale. Funding in the space is significant: Deepnote raised $23.8M (Index Ventures, Accel, Y Combinator) and Databricks is valued at over $62B as of 2024. Primary buyers are data science and ML teams at mid-size to large companies, university research groups, data engineers building reproducible pipelines, and analysts who want notebooks without the infrastructure burden.

Project Status

This project is in the research and specification phase.
Contributions, feedback, and domain expertise are welcome.

Contributing

We welcome contributions from developers, domain experts, and potential users. See CONTRIBUTING.md for guidelines.

Important: All contributions must be your own original work or clearly attributed open-source material with a compatible licence. Copyright infringement and licence violations will not be tolerated and will result in immediate removal of the offending contribution. If you are unsure whether a piece of code, text, or other material is safe to contribute, open an issue and ask before submitting.

Licence

Licence to be determined. See discussion for context.

core