Deepnote 2020: The Year in Review

We are building a new computational medium for data science. This is a look back at key milestones of 2020 and what we’ve learned along the way.

When we started to work on Deepnote, we knew it wouldn’t be easy. We had to solve hard technical problems. We had to invent new interfaces. We had to build new tools — tools that help us to reason, experiment, and collaborate.

Despite all the challenges that the year brought, 2020 was big for us at Deepnote. We told the world about our mission for the first time. We showed the community what we are building. First in private beta, then in a public beta. And we went from 0 to thousands of users. All thanks to the amazing team and to all our early users who joined us on the journey.

This year, our team grew to 15 engineers, designers, and researchers. We are all over the world, but mostly in San Francisco and Prague. We’ve shipped countless major features and small delights. Some were obvious, some were unintuitive. We’ve focused on 3 key themes:

1. Collaboration — We deeply believe in the collaborative aspect of data science so we build Deepnote as a tool where collaboration is a first-class citizen. We designed support for teams, comments, and versioning to be a natural part of the workflow.
2. Beautiful interface — Our vision is to create a computational interface that is powerful for data scientists, yet easy to use for all. We worked hard on building an intuitive interface that’s easy to control via a powerful command palette and keyboard shortcuts.
3. Integrations — We heavily invested in integrations so that Deepnote plays well with existing tools, processes, and the rest of your data stack.

Without further ado, here’s a recap of how the Deepnote product, team, and community evolved in 2020.

January: Collaborative features

Deepnote is a collaboration-native tool. We already had real-time multiplayer editing in place, but that by itself is not enough. When people collaborate, they don’t use just tools, they also define processes. We had already seen how our users are exchanging messages by writing Python comments. Now it was time to step up our game and in January we introduced comments. You can use comments to give and respond to feedback, track changes, suggest improvements, and, in general, iterate faster.

February: Telling the world about us

In February, we announced that we raised $3.8 million seed round and welcomed some amazing investors on the board, including Index Ventures, Accel, Y Combinator, and Credo Ventures, as well as a number of angel investors, including Greg Brockman, Dylan Field, Elad Gil, Naval Ravikant, Daniel Gross and Lachy Groom. This was the first that we told the world what we are building.

March: Code intelligence

We introduced our first code intelligence features. Even though data science workflows are fundamentally different from software engineering workflows, there’s a lot we can learn from decades of research in programming languages and developer tooling. Deepnote shows you function documentation and autocompletes your parameter names so you can focus on the problem, not the code. Since March, we’ve also added a linter that analyzes your code to flag stylistic errors, bugs, and suspicious constructs. Deepnote now has full code intelligence as you may know it from IDEs.

April: Command palette

How do you build a powerful yet intuitive interface? Command palette in Deepnote gives you superpowers with a single shortcut. It provides quick access to all your files and most popular actions (e.g. run all, delete a cell, hide code). Just press Cmd + P on a Mac or Ctrl + P on Windows and start typing.

May: Teams

In May, we introduced Teams, a key milestone in improving the collaborative experience. With Teams, you can create a shared space for all your collaborators and share projects, files, and environment configurations. You can also choose different access levels for maximum security, just like you would in a GitHub organization or in a shared folder on Google Drive.

June: Deepnote community

Our community started sharing what they’ve been building using Deepnote. And there are some awesome things — like a course teaching you NLP the Stanford way, an overview of most-used Python packages, or a showcase of how to create your own data sets. If you want to join our community of data scientists and makers and share your work with the world, come and say hi.

July: Cell outputs export

In July, we introduced cell outputs export — every cell comes with a unique link that you can share with others or embed into tools like Notion. Shared outputs update whenever the original cell is executed, so you can create live dashboards.

August: Integrations

In August, we doubled down on integrations. First, we introduced GitHub & GitLab integration, allowing you to link your Deepnote project with your GitHub repository and commit in one click. We’ve also added BigQuery, AWS S3 buckets, MongoDB, Snowflake, PostgreSQL, and the list of integrations is still growing.

September: Cell management

Even though cells are not a new concept, they’ve been traditionally underutilized and their benefits generally remained unexplored. We experimented with ways how to structure and navigate your codebase, as well as direct the attention of readers to relevant parts of the notebook. We added support for hiding the output of a cell or the code itself. This is a collaborative feature, so if you are presenting your notebook to someone, you can easily hide the code and only show them the relevant outputs.

October: Public beta

In October, we’ve opened up our private beta to all and the support has been overwhelming. On launch day, we got trending on Hacker News, Product Hunt, and Reddit. In the first month, we’ve 5x-ed our daily active users and sign-ups. Thank you for helping us spread the word!

November: Versioning

Versioning has been one of the most requested features since day 1. In the first iteration, we introduced history which allowed you to see the list of actions by your collaborators. Today, you can see the previous states of your notebook and revert to past versions, either from snapshots you create yourself or the ones we create automatically for you. This helps you collaborate more effectively and should make for a smoother experience in your code reviews.

December: First-class support for SQL

We believe SQL is a first-class citizen of data science, so in December we added a new kind of cell into the notebook — a SQL cell. With SQL cells, Deepnote highlights your syntax, suggests table and column names when typing, and outputs pandas data frames, Deepnote-style.

As a small bonus, we’ve pre-installed a ton of libraries into every new Deepnote project, so now you have your favorite data analytics and ML libraries ready in Deepnote, including numpy, pandas, tensorflow, keras, nltk, seaborn, and more. All you have to do is import them into your project.

What’s next

2020 has been a busy year, but there’s so much ahead. In 2021, we will continue to deliver the best possible notebook experience with stronger machines and explore more advanced concepts from exploratory programming. We also want to improve your experience of working with non-data scientists, add in layers of interactivity, and grow beyond literate programming with a new set of cell types. If this sounds too exciting to just stand by and watch, we’re always looking for smart engineers. Join us.

Thank you for sticking with us, supporting us, and being a part of our awesome community. Here’s to more adventures in 2021! 💙