Data science beyond data science teams

Two years ago, we started Deepnote to build the best data science notebook. We wanted to develop the solution we desperately wanted to use ourselves — something that would allow us to work in notebooks effectively, in the cloud, with all the bells and whistles we were used to from the software engineering world.

But data science goes beyond data scientists. We’re building not just the best notebook for data science teams. We are building the best notebook for teams working with data.

Today, we’re launching Deepnote for Teams. And going forward, we will be building even more features that allow anyone in any organization (not just data scientists) to have a seat at the table.

When starting Deepnote, our initial focus was to build a notebook that would support data scientists. We allowed them to work together on the same projects at the same time, added all integrations we could think of, embedded code intelligence so that data scientists can use Deepnote like they’d use their favorite IDE, cracked the many challenges around versioning and reproducibility. We believe we have delivered on that initial vision, but to make data endeavors truly successful, a lot more is needed. Data science is as much a scientific and a creative process as an engineering one. It involves failing, learning and going back to the drawing board. It requires a large amount of communication with other teams and stakeholders.

Data science is bigger than data scientists. It requires:

• Framing the problem with business stakeholders
• Finding the right, up-to-date, clean data with data engineers
• Continuous conversations and iterations with product owners
• Analyzing the prototypes, collecting feedback, and iterating on models among the data science team
• Sharing the findings as dashboards and reports with customers, suppliers, or investors
• Deploying the models to production with the help of the engineering and devops teams

Our next horizon is building a product that is powerful for data scientists, but at the same time, inclusive and accessible enough for everyone else in the organization.

Reimagining the data science process

Before we dive deeper, let’s take a step back and look at the market trends that are underpinning this shift.

Trend #1 — The rise of the citizen data scientist

Using browser-first collaborative tools is becoming the new norm. Google Docs, Notion, and Figma paved the way. And much like the design process changed with the arrival of Figma, the data science process is going through the same fundamental transition. Traditionally, data scientists used to work in isolation, only sharing the outcomes on their work once the process has been completed. With increasing availability of data, there’s a lot more curiosity on how to use it from all across the organization.

In the organizations of tomorrow, data scientists work directly with developers, product managers, or marketers around the same analysis, on the same data set, with different access rights. This leads to shorter iteration cycles, faster releases, better outcomes and more engaged teams. Even better, the data scientists are developers, product managers, and marketers.

To support this transition and empower citizen data scientists, we need tools that allow anyone to access data without barriers. But accessibility without understanding is not useful. We also need tools that help anyone in the organization analyze the data and apply insights autonomously.

Trend #2 — Competing in a data-driven world

In line with that, the second trend we’re seeing is organizations realizing how data drives value and scaling their data capabilities to leverage that. The gap between industry leaders and laggards driven by organizations’ use of data and analytics keeps growing, and getting ahead of the curve is the core focus in many data-driven organizations.

As one of our customers, an Engineering lead in a life sciences company said, “I believe growing data science capabilities across the team today will give us a competitive edge later. I want for all of our scientists adopt these skills, so I like bringing everyone directly into notebooks.”

This is the case in life sciences as much as in retail, insurance, consulting, manufacturing and other legacy industries. Companies with the greatest overall growth in revenue and earnings receive a significant proportion of that boost from data and analytics. Building data science capabilities is what can get you the competitive edge.

Trend #3 — Remote-first workforce

With COVID, our work lives are moving remote. A McKinsey survey found that 90 percent of executives envision a future that’s either remote-first or hybrid, with some combination of remote and on-site work. This model of collaboration brings new challenges and requires new tools to step up and help. We need interfaces that allow us to collaborate effectively in any setting.

So how does Deepnote fit in?

At Deepnote, we believe notebooks are the tool that can support these trends and help companies make a shift from data scientists working in isolation to a data science process that is decentralized and inclusive for all in the organization. We want to help data-driven teams explore, analyze and present data from start to finish. How do we do that?

1. Notebooks paradigm

First of all, Deepnote is a notebook. Notebooks were designed for exploratory programming and rapid prototyping. They were designed for answering questions. We love the notebook paradigm because we think it’s really useful for a productive conversation around the data. Teams see the data source and the logic in one place, they can rapidly explore and visualize data and iterate on their findings. Naturally, this makes notebooks a great entry point for non-technical collaborators, who can easily understand the context, and with Deepnote, contribute to the conversations using features like comments, no-code visualizations, and input cells where they can change variables without code.

2. Browser-first

Deepnote is browser-first, allowing everyone to spin up a notebook in a couple of seconds with no hardware provisioning, no installation, no engineering support. “We’d had a lot of technical issues when trying to pair up on a Jupyter notebook remotely. Deepnote is incredibly easy to set up and allows us to start new notebooks in seconds,” shared Becca Carter, Product analytics lead at Gusto. Having a browser-first notebook also means that everyone interacts in the same environment, with the same version of the notebook. Which brings us on to…

3. Collaboration-native

Deepnote brings team together and allows them to work together, either in the real time or asynchronously. Luca Naef, CTO of VantAI who uses Deepnote for real-time code reviews with his team summarized this perfectly. “Having direct access to the runtime and program state which makes understanding complex models much easier and leads to much more spontaneous creative ideas.”

Where are we going?

Ultimately, our vision for Deepnote is to helps teams explore, analyze and present data from start to finish. No more silos — we want to make notebooks the focal point for any data team.

In the coming weeks, we will be rolling out new features that will help us deliver on this vision. No-code visualizations so that can explore your datasets hands-on, interactive apps and dashboards, and publishing.

If you want to try out what a collaborative data science workspace feels like, we’ve just launched a free plan that allows you test to the full collaborative powers of Deepnote. Also, if you’re excited about our mission of building a team workspace for data, we’re hiring!