Sign inGet started
← Back to all posts

Revolutionizing data work with Deepnote AI - again

By Megan Lieu

Updated on October 26, 2023

If you thought we were done reshaping data workflows with AI, we’ve been working on one more addition to the Deepnote AI family. And guess what? It’s the best one yet.

Illustrative image for blog post

The first step in this journey to transform data work in notebooks via the power of AI was through our AI Copilot, which we launched back in June this year.

Since then, we have received plenty of feedback on how to get even more out of combining conversational AI with notebooks as a computational medium. And we listened.

After code completion, we tackled code generation, editing, and debugging using generative AI, then brought that to the next step of multi-block generation. That journey has culminated in our most ambitious project yet—a completely autonomous, self-sufficient agent called Deepnote Auto AI.

Combining AI and notebooks

When you picture the impact of AI in your everyday life, you might picture AI chatbots or even AI assistants that can supplement your day-to-day tasks. But we challenge you to think bigger: an AI collaborator who doesn’t just provide suggested ways to do things, but also has autonomy to execute tasks and come back with the answers you seek.

We’re banking on virtual collaborators becoming the next big thing for analytical AI. To a certain degree, it’s already working in the form of ChatGPT Advanced Analysis: you can ask it to do some analysis, it will write code, execute it, evaluate the outputs, correct itself and at the end provide you with some final output, like an image of a chart.

We're convinced that notebooks, in contrast to chat windows, are the ideal platform for interactions with generative AI. Their structured, block-by-block format offers a transparent and reproducible framework for the functioning of AI agents. Add to this Deepnote's collaborative workspaces, robust data connectors, and flexible environments, and the result is a vastly superior experience.

Built with the citizen data scientist in mind

More and more people who are doing the work of a data scientist are those without the title of data scientist. Data teams are becoming less siloed and more embedded throughout their organizations. These trends are contributing to the push for more collaborative data work and for tools that aid in bringing everyone to the table when it comes to data-driven decisions.

Additionally, the future of generative AI in the data landscape is not just about using it to improve productivity of seasoned data professionals who work on advanced use cases, but also to break down barriers so that professionals of all levels and titles can do data science and analytics.

This is the context upon which we developed Deepnote Auto AI. Anyone who wants to or needs to contribute to data workflows in their organization, regardless of their technical skillset, can open a new notebook, provide some basic instruction to Auto AI, and generate an entire notebook—complete Python code to transform and visualize the data, develop predictive models, and text narration throughout the notebook to tie all the findings together into one cohesive story—with a few clicks of a button.

By enabling more and more people to adeptly create these end-to-end data projects, Auto AI is our answer to unlocking a world of possibilities when we bring together more bright minds to the data table.

Explore the possibilities

Auto AI is now available within the comfort of the AI button you’ve come to know and love. When you turn on the Autonomous AI toggle within this modal, your AI teammate not only generates multiple code blocks for you based on your instruction, but also goes on to execute them for you.


But Auto AI’s intelligence goes well beyond just its code generation capabilities.

Uses context for smarter recommendations. Auto AI uses any context you have provided, such as text blocks describing the task at hand and variables stored in memory to inform its suggestions. Coming soon, Auto AI will also be able to learn from your workspace, integration schemas and file systems so that it generates code with just as much knowledge of your data systems and organizational structures as you have.

Builds upon prior outputs. Each subsequent block generated uses context from preceding blocks in the notebook, including ones that it generated itself within the same continuous run. Access to those outputs provides additional information, enabling Auto AI to assist you at any stage of your analytical journey. You can make it start the analysis for you from scratch, or you can ask it to help out when you get stuck somewhere in the middle.

An added asset to your team. You can be confident in not only Auto AI’s outputs, but also that a colleague can pick up where you and the AI agent left off. Auto AI ensures transparency and reproducibility by documenting the original prompt used as the basis for the auto-generated notebook, as well as the generated code and outputs side-by-side, acting as the glue to enable true collaboration between you and your teammates.

Works while you sleep. Auto AI is that hard working coworker of yours who never seems to need any sleep. Need to step away or just navigate to another notebook? Rest assured your AI companion is still hard at work in the background, and will probably finish the assigned task faster than it takes you to context switch back. In the meantime, easily monitor the progress of your Agent’s work with a new status bar available at the top of your notebook.


Completely autonomous and self-sufficient, with control. Auto AI is smart enough to start up your machine, fix its own errors and basically do anything else it needs to get an entire notebook over the finish line without any further instruction. But if at any point you feel the need to interrupt or change its course, you have full control to stop the run.

End-to-end AI assistance

With the addition of Auto AI to the Deepnote AI suite of tools—including AI Copilot, single- and multi-block code generation, editing, explaining and fixing—we’ve come full circle in making AI a first-class citizen in Deepnote. From the code level all the way up to the entire notebook level, you can now use AI to enhance your productivity from beginning to end of your data science workflows.

With this variety of AI capabilities now available in Deepnote, you can customize your workflows to fit your preference in AI assistance level, with AI Copilot giving you the lightest boost all the way to Auto AI providing the most assistance.

Check out our docs to learn more.

Watch it in action

Try it out yourself and let us know what you think

Auto AI is now available to all Team and Enterprise customers starting today, but anyone can try out Deepnote’s full set of AI capabilities for free on a Trial plan.

But this is not the end of the AI journey. We’re committed to making our AI tools even more powerful—expect even more contextual knowledge and additional tools in Auto AI’s arsenal coming soon!

This journey for us at Deepnote would not have been possible without your feedback along the way. Please continue sharing your experiences with us in our Product Portal.

Autonomous AI notebooks are by far the most aspirational endeavor we’ve set out to create, and we’re confident that with your continued feedback over the next few weeks as you get your hands dirty with it, we can shape the notebook of the future together.

Megan Lieu

Data Advocate @ Deepnote


Illustrative image for blog post

Beyond AI chatbots: how we tripled engagement with Deepnote AI

By Gabor Szalai

Updated on April 3, 2024

That’s it, time to try Deepnote

Get started – it’s free
Book a demo



  • Integrations
  • Pricing
  • Documentation
  • Changelog
  • Security




  • Privacy
  • Terms

© Deepnote