# deepnote-jupyter-successor-ai-first-github-2025-11-07

## Veille

Deepnote - Jupyter Successor with Native AI Agent - .deepnote YAML Format - VS Code/Cursor/Windsurf Extensions - Open Source - GitHub 1.5k stars

## Titre Article

Deepnote: the data notebook for the AI era

## Date

2025-11-07

## URL

https://github.com/deepnote/deepnote

## Keywords

Deepnote, Jupyter, notebooks, data science, AI agent, format .deepnote, YAML, open source, VS Code extension, Cursor, Windsurf, JupyterLab, block-based architecture, reactive execution, version control, real-time collaboration, Python, R, SQL, data visualization, @deepnote/blocks, @deepnote/convert

## Authors

Deepnote Team (Johnny Carrot et contributeurs)

## Ton

**Profile:** Open-source product | Organizational first-person collective | Informative-promotional | Intermediate

The Deepnote team adopts an open-source product marketing voice, positioning the solution as a "Jupyter successor" in the AI era. The comparison-table-plus-value-propositions structure (human-readable format, block-based architecture, "work wherever") reflects a product positioning strategy. The data science practitioner language (notebooks, reactive execution, kernel compatibility) targets data professionals and research teams. The tone is educational-promotional, typical of open-source projects, balancing technical depth and accessibility. The emphasis on collaboration and AI-native design reflects the priorities of modern data science workflows. The acknowledgements section toward the Jupyter community frames the innovation within a respectful narrative. Typical of venture-capital-funded open-source startups (Observable, Hex, Mode) building next-generation data tools for teams seeking a modern collaborative notebook experience.

## Pense-betes

- **Jupyter successor** with AI-first design and compatible kernel
- **500,000+ data professionals** as users (Estée Lauder, SoundCloud, Statsig, Gusto)
- **.deepnote YAML format** replaces .ipynb JSON - human-readable, version-control friendly
- **Block-based architecture**: SQL, inputs, charts beyond classic code cells
- **Reactive notebook execution**: automatic re-runs of dependent blocks
- **Available extensions**: VS Code, Cursor, Windsurf, JupyterLab
- **@deepnote/blocks**: TypeScript types for blocks (Code, SQL, Text, Markdown, Input, Viz, Button, Big Number, Image)
- **@deepnote/convert**: CLI tool for .ipynb ↔ .deepnote conversion
- **Deepnote Cloud**: team scaling with cloud compute, real-time collaboration, AI agent
- **Deepnote Open Source**: local work before scaling to cloud
- **1.5k GitHub stars** - Apache-2.0 license - 14 contributors
- **Free for academics**: unlimited access for students and teachers
- **Roadmap**: local UI, local AI agent, bring your own keys, run your own compute
- **Multi-language support**: Python, R, SQL
- **Built on Jupyter kernel**: full compatibility with existing notebooks
- **Native integrations**: built-in database & API connections

## RésuméDe400mots

Deepnote is an open-source project positioned as a modern successor to Jupyter, designed for the era of artificial intelligence. Used by more than 500,000 data professionals at companies such as Estée Lauder, SoundCloud, and Statsig, Deepnote combines Jupyter compatibility with AI-native functionality and an advanced collaborative experience.

**Format Innovation**

The `.deepnote` format replaces the verbose JSON of `.ipynb` with a human-readable YAML structure optimized for version control. This format organizes multiple notebooks, integrations, and settings within a single .deepnote project, facilitating structure and collaboration. The `@deepnote/convert` CLI tool enables seamless bidirectional conversion between Jupyter and Deepnote formats.

**Extensible Architecture**

Deepnote introduces a block-based architecture beyond traditional code cells. Through the `@deepnote/blocks` package, users access blocks for SQL queries, interactive inputs, visualizations, buttons, big numbers, images, and separators. These blocks are defined and validated in TypeScript, providing type safety and extensibility. Reactive notebook execution ensures dependent blocks automatically re-run when inputs or data change, maintaining consistency and reproducibility.

**Multi-IDE Ecosystem**

The open-source project provides official extensions for modern AI-native editors: VS Code, Cursor, Windsurf, and JupyterLab. This "work wherever" strategy allows data scientists to develop locally in their preferred IDE, then scale to Deepnote Cloud for real-time collaboration with robust cloud compute.

**Hybrid Cloud-Local Strategy**

Deepnote Open Source enables complete local development, while Deepnote Cloud offers managed compute, a native AI agent, link-based sharing, native database/API integrations, and synchronous collaboration. This hybrid approach addresses the needs of individual data scientists (local, free, full control) and teams (collaboration, scalable compute, governance).

**Roadmap and Vision**

The roadmap includes the local Deepnote Cloud UI, a local AI agent, bring-your-own-keys support for AI services, and run-your-own-compute capability. These developments aim to offer the full Deepnote Cloud experience locally for users requiring data sovereignty or working with sensitive data.

**Positioning vs. Jupyter**

The comparison table highlights zero setup (cloud or local vs. local installation), native AI features (agent and code completion vs. third-party extensions), integrated version control (native Git vs. manual workflow), simplified sharing (link vs. manual export), managed compute (cloud vs. local resources only), and native integrations (databases/APIs vs. manual configuration).

**Academic Community**

Deepnote Cloud is free for students and teachers, with unlimited access to core features, cloud compute, and real-time collaboration. The project encourages academic citations and contributes to the open-source data science ecosystem.

**Jupyter Legacy**

The acknowledgements pay tribute to the Jupyter community and its impact since 2013, positioning Deepnote as a natural extension of this legacy toward an AI-native, collaborative future, while actively contributing to the same open ecosystem.

## GrapheDeConnaissance

- Deepnote —remplace→ Jupyter (TECHNOLOGIE, 0.95)
- Deepnote —utilise→ format .deepnote (TECHNOLOGIE, 0.98)
- format .deepnote —remplace→ format .ipynb (TECHNOLOGIE, 0.97)
- Deepnote —est_basé_sur→ Jupyter (TECHNOLOGIE, 0.97)
- Deepnote —utilise→ AI agent natif (TECHNOLOGIE, 0.95)
- Deepnote —mesure→ 500 000 utilisateurs (MESURE, 0.92)
- Deepnote —s_applique_à→ VS Code (TECHNOLOGIE, 0.98)
- Deepnote —s_applique_à→ Cursor (TECHNOLOGIE, 0.98)
- Deepnote —s_applique_à→ Windsurf (TECHNOLOGIE, 0.98)
- @deepnote/blocks —permet→ types de blocs notebooks (CONCEPT, 0.93)
- @deepnote/convert —permet→ conversion bidirectionnelle de notebooks (CONCEPT, 0.95)
- Deepnote —utilise→ exécution réactive (CONCEPT, 0.9)
- Johnny Carrot —collabore_avec→ Deepnote (ORGANISATION, 0.85)
- Deepnote Cloud —permet→ collaboration temps réel (CONCEPT, 0.97)
- Deepnote —utilise→ licence Apache 2.0 (CONCEPT, 0.99)

---
Canonical: https://www.thekb.eu/en/fiches/deepnote-jupyter-successor-ai-first-github-2025-11-07/