Learn about Hex, a collaborative data analysis platform, and the future of data apps built on top of the data warehouse
I had the pleasure of getting early access to Hex over the past few weeks and it's starting to shape a lot of my thinking about what the future of data apps might look like, and how data in general can start to learn more from the field of design.
On the surface, Hex is a collaborative data analysis platform that enables data people to ingest, explore, and visualize data using both Python and SQL. It may not appear that different from tools like Jupyter Notebook, RStudio's excellent IDE, or even Shiny apps, but the more time I spend with it, the more I think there's an exciting potential here.
Collaboration in Hex is not just an add-on, but a core part of the product functionality. In many ways, you can see the link between the success of a tool like Figma, and the design philosophy of Hex. Kevin Kwok's excellent view on why Figma was such a successful product can help us better understand why Hex might win too.
Hex enables an iterative workflow where you can design data applications for use by your stakeholders. Being able to share the app in development with your colleagues means that you can test your assumptions about what makes sense and what doesn't instantly.
Let's dig through Hex, with the caveat that Hex is in active development so some of this might have changed since I've written it.
To help make this overview a little more fun, let's imagine a scenario. We've noticed our customer success team is often spending a lot of time writing queries and asking for help when trying to better understand our customers.
We've decided to pair with them on a project to help make the common questions they ask more readily available. After spending some time with them, we've come up with a few recurring scenarios:
Our success team:
We could solve this using SQL that we send them to run, but we think an app might be a better experience for everyone involved, so we're going to build our first Hex App for them.
The latest version of Hex offers a reactive analysis format. If you're used to traditional notebooks like Jupyter or R Studio, you know that your only option for running cells is from top-to-bottom or one-at-a-time, but has has built an internal DAG of your analysis and can infer the dependencies of one cell downstream to all other cells, meaning you can ask Hex to only run cells that would've changed in response to a change in your upstream inputs. For our example, we'll use the Reactive mode, because it's way cooler (and I managed to get early-access)
The first thing we'll do is update the name on the top left so it's no longer called Untitled Project. For now, Customer Success demo will do.