From working with external developers to providing machine-learning capabilities, Snowflake co-founder Benoît Dageville says the firm is focused on honing its own data capabilities while also providing a platform for best-of-breed solutions that will help customers solve fresh challenges in fast-evolving areas.
Dageville spoke with diginomica at Snowflake Summit 2022 in Las Vegas, where the company announced a slew of new products, including data workloads, app development frameworks and security features.
These announcements form part of the company’s continuing efforts to develop what it refers to as the Data Cloud, which aims to create an integrated and cross-cloud platform to eliminate data silos. Dageville says his product team is now concentrating on ways to extend this platform:
Every two or three years, we are completely opening and broadening the definition of Snowflake. At the same time, we are still working really hard on data warehousing and Big Data and data lakes and all these other technologies. So, I would say the hard work is never over. But we have broadened what Snowflake means. We don't want to be a small part of the stack. We have broad ambitions.
Those innovations are important because evidence suggests many organizations still struggle to exploit the vast troves of information they hold. Dageville said his company continues to develop its own data-sharing capabilities and, crucially, is allowing external developers to build and run their own applications on the Snowflake platform:
The ecosystem partners are super-important for Snowflake. We are really open in many aspects and very inclusive. Snowflake will be mostly not be built by our employees – that's the vision. That's why we are being flexible in terms of running other developers’ applications.
The company is drawing on a wider ecosystem of developers to help customers deal with key data concerns, such as cybersecurity. By adding best-of-breed solutions to its own technology platform, Snowflake wants to help executives find novel solutions to their business challenges. So is the intention to create a ‘one-stop-shop’ for the data stack by blending its own tech with external partners? Dageville says:
One distinct attribute of our Data Cloud is that, instead of building a completely disconnected product and solution that you have to assemble and make work together, we decided for the reason of simplicity to create one consistent product, one single product. And this is very important. We might create more than one product one day. And there will always be applications built on top of that consistent, single product. But for now, the entire data stack is one single product.
This consistency is significant because it ensures processing effort is honed on the right areas. Dageville gave the example of ML capabilities, suggesting that as much as 80% of efforts in this area are connected to data pipelines and feature engineering, and just 20% to actual data processing. Any company that wants to serve ML efficiently will need to ensure the underlying capabilities are strong and connected, he argues:
Putting everything under one single product removes friction. And Snowflake is all about removing friction and democratising things, so that everyone can be an application builder. And that's what we are really after.
What’s coming next?
In terms of forthcoming innovations, Dageville says he always tries to think a few years ahead when it comes to developing fresh products. He said the most likely way that organisations will be able to make more of their data is working with other like-minded business. Dageville expects Snowflake to focus on this area even more during the next few years through the creation of an app store:
As we started thinking about collaboration, it was obvious to me that applications are about more than just data – you want to collaborate. This is where Snowflake is going to be an amazing place if we can build this app store of data applications. This store will bring companies together. It's not a consumer store, like on your cell phone, but it’s the same idea. And this is where we can democratize data.
To help build this app store-like approach, Snowflake announced a new Native Application Framework at Summit that will allow developers to build applications and monetise them on Snowflake Marketplace, which is a collaboration space that allows companies to exploit a variety of third-party datasets. The next phase of development for Snowflake involves helping companies explore ML capabilities and create a means to share this new expertise:
ML and data applications is where we are going after and, obviously, we have a long, long way to go. We are just at the beginning of that journey. And there is unstructured data and deep learning and complex frameworks to run in Snowflake. If you want to know where we are going, it’s really to be this ML platform, but not only where you will build and use it yourself. A lot of people are going to do that probably, but the revolution is the sharing of ML and the sharing of this expertise.
Snowflake’s broad focus on collaboration has been accompanied by an attempt to make it much easier for companies to share knowledge. In his keynote speech, Snowflake CEO Frank Slootman explained how companies in a range of industries – such as retail, healthcare and life sciences – are using the Data Cloud to share insight. Dageville says such verticalization is central to the company’s long-term strategy:
We want to make it really easy for these different industries to connect to the cloud and to connect the expertise. Financial companies, for example, need to connect with other financial firms because they have sector-specific problems to solve. We are very aligned with our customers and, therefore, we are very vertical in our way of thinking.
Lots to come. Meanwhile many of Snowflake’s newly announced features – such as Unistore, which is a workload approach that allows companies to work with transactional and analytical data together on a single platform – are aimed squarely at continuing efforts to break down data silos. As Dageville concludes:
Making it seamless to get all your data into Snowflake is really important for us. And, of course, that’s easier said than done. Because moving data and un-siloing data is hard. So, we are trying really, really hard and it’s a huge focus for us. Data is our number one focus, by far, because it all starts there. Nothing can exist without data.