Sully McConnell is the Head of Insurance at Snowflake, delivering the AI Data Cloud to help organizations share data, build apps and power their business with AI. Sully was interviewed by Chris Luiz, Director of Solutions Architecture and Customer Success at Monitaur and Cleveland Founder at InsurTech Ohio.
Sully, Snowflake is known as a foundational data platform, but it’s also a data-sharing platform. Can you describe how Snowflake is enabling customers to share data?
“When Snowflake built this platform 10 years ago, the innovation was that if you were going to build a platform for the cloud, you would architect it very differently. The first big architectural decision that Snowflake made was the idea of separating compute from storage. Meaning in the cloud, you can store privacy-protected data once, but because it’s separated from compute, different independent compute clusters can connect to that single copy of data.
That has many positive implications for our customers. In each of those independent compute clusters, you can run different workloads. I can run my AI/ML workloads, my dashboard / scorecard workloads, my engineering workloads, etc.
I can do near real-time updates to this single copy of data, all independent of each other. There's no conflict for machine resources, so that's all fantastic. The bigger thing is that the architecture works across Snowflake account boundaries. If I have data in my Snowflake account and it's interesting to a business partner of mine, I can very easily provide them a view of that single copy of data.
With just a few clicks inside my Snowflake account, a view of my data would appear in my business partner's Snowflake account, query-ready and join-ready. This secure data-sharing across account boundaries simplifies each organization's operations and opens up interesting use cases where multi-party data can have an impact.”
Data is the foundation of so much of what we do and build these days. Given the ease of data sharing on Snowflake, what does that enable downstream?
“I'm going to describe two common kinds of data sharing and the downstream implications. The first is the idea of a third-party data marketplace. Again, because it’s relatively easy to share across account boundaries, we now have a marketplace, which people would say is the most impactful one in the industry.
As of October, there are over 2,900 third-party data sources in the marketplace, but that grows more each month. If there's a third-party data provider that wants to have a relationship with you, in a couple of clicks, they can expose that third-party data in your Snowflake account. It's still a single copy of data, but now it appears in your Snowflake account and is joinable to your first-party data.
The second is simple as well. Imagine as an insurance carrier, you're using a prominent SaaS application for Claims, Policy-Administration or Billing. As you use those systems transactionally, vendors will write an analytic data set into Snowflake and share it with you. The benefit of that is you don't have to go through an enormous integration effort. Typically, organizations that use those vendors have a big integration effort to get data out of those platforms and into their data warehouse, data marts and analytical systems. That gets radically simplified with vendors leveraging Snowflake's data sharing capabilities.
Again, you wake up the next day, and in your Snowflake account, there are many tables with your transactional data. It's joinable to your other first-party data. An additional benefit is having this remarkable set of third-party data you can use for numerous use cases. It simplifies operations either for the third-party data or your own transactional data from the SaaS vendors that are sharing data via Snowflake as well.”
What are some of the benefits your customers are realizing today from this functionality and these applications?
“There are technical benefits and business benefits to touch on. Technically, data sharing simplifies operations. You don't have to manage a thousand jobs, bringing data in from different transactional systems. If you can implement sharing with a vendor partner, you don't have to run all of the jobs and have all of the ETL (Extraction, Transformation, Loading) and integration logic. You get reduced implementation costs because it's easier to get simplified operations.
You can build more sophisticated pricing models and ensure more accurate underwriting on the customer side. You can get better customer insights and connect your first-party data to third-party data firmographics or demographics, so you have a better profile of a person or a commercial business to do better targeted marketing. You can use weather data, IOT-sensor data and satellite imagery.
On the claims side, you can leverage that to streamline claims processing and get more accurate claims assessments. The use cases that evolve out of this idea of sharing are fantastic in the industry.”
What are you most excited about over the next 12 months?
“It’s an interesting time in the Snowflake ecosystem. What I'm most excited about is building on this idea of sharing. The industry thinks of Snowflake as a great platform for data sharing, and that’s true, but that's a little bit of a misnomer. Snowflake has an architecture that supports this idea of securely and easily sharing across Snowflake and account boundaries. Those things can be data sets. Right now, you're seeing solution providers saying, ‘I'm going to build an application that could run inside of Snowflake and provide that to consumers of Snowflake from their point of view.’
We're seeing solution providers build full blown applications inside of Snowflake, both horizontal and industry vertical applications. Examples would include geospatial applications, graph analytics applications, marketing and media clean rooms, time-series applications and large language models, just to scratch the surface. A common industry problem is the need to resolve someone’s identity who’s been interacting across multiple touchpoints of an organization. There are applications inside of Snowflake that help with identity resolution while leveraging their proprietary third-party data to provide enrichment. There are also vision model applications. I can take aerial imagery of a house and extract features like rooftop condition or geometry. These are the applications being built from the builder’s point of view. The provider of these solutions gets to approach customers with interesting options.
‘Would you like to try this thing?’ In a couple of clicks in their Snowflake account, you can enable that application to show up in the customer's account. It can be applied to the customer data. There's no setting up of infrastructures or installation of softwares. The data never leaves the customer's account, so you don't have to worry about copies of privacy-protected or sensitive information floating around in multiple sets of infrastructure. It becomes easy for the vendor to offer applications. If the application is interesting to the customer, deployment is already effectively complete
The implication of Marketplace Applications is that a customer has access to numerous applications and can try them quickly and easily. There's this idea that rather than buying big, monolithic things, you can spend less money and purchase targeted point solutions. From the consumer point-of-view, you're going to get a quicker evaluation, speed to market, speed to insight, no copying data around the enterprise, fewer copies of PII or sensitive information, a reduction in redundant infrastructure, easier to manage privacy regulations, etc. This is the reason I'm excited. This is going to be game-changing for customers to take advantage of this in the environment.”