Despite the seemingly nonstop conversation surrounding AI, the data suggests that bringing AI into enterprises is still easier said than done. There’s so much potential and plenty of value to be captured — if you have the right models and tools.

Implementing advanced AI today requires a solid data foundation as well as a set of solutions, each demanding its own tools and complex infrastructure. That means moving data from one point solution to another and then back to its home, which can be a costly, clunky and not terribly secure process. 

But RelationalAI saw a different paradigm enabled by the emergence of the AI Data Cloud: creating tools that can work where the data lives, inside a company’s Snowflake account, so every team, group leader, organization and executive can combine the data they have in Snowflake with knowledge about the business to make better-informed decisions.

By adding support for graph, rule-based reasoning, prescriptive and predictive analytics directly in Snowflake, “we help Snowflake deliver on the promise of ‘all data, all workloads,’” and make complex decision-making techniques more accessible, says Molham Aref, CEO of RelationalAI. 

Snowflake Native Apps and Snowpark Container Services: Better together

RelationalAI’s Snowflake Native App leverages Snowpark Container Services (currently in public preview) and runs fully embedded within Snowflake as a relational knowledge graph coprocessor. Knowledge graphs present a digital model of an organization’s operations, surfacing patterns, relationships and connections that RelationalAI’s graph algorithms use to detect similarities and apply reasoning and business logic. 

A fully managed service, Snowpark Container Services is designed to facilitate the deployment, management and scaling of containerized applications — all within the Snowflake environment. Whether it’s security or configuration, Snowpark Container Services handles the intricacies, allowing users to instead focus on their applications without the overhead of managing the underlying infrastructure.

“[Snowpark Container Services] is great as a building block,” says John Macintyre, Vice President of Product at RelationalAI, “but it's really the Snowflake Native App Framework that ties it all together for customers to provide an integrated experience.”

By bringing workloads closer to the data, Snowflake Native Apps integrated with Snowpark Container Services makes it easier for RAI’s customers to adopt its technology. 

“Historically, because of security and procurement reviews, it could take several months to get users in enterprise organizations access to the product," Macintyre explains. “We've seen with Snowflake Native Apps and Snowpark Container Services, that can now happen in days and even hours, which is mind-blowing.”

That speed to value, plus the elasticity and power of RelationalAI’s knowledge graphs, makes it easier for companies to run deeper analytics, faster. For example: AT&T applies RelationalAI’s cloud-native knowledge graph solution to its data and runs graph analytics to primarily detect fraud and insider threats, as noted by AT&T Director of Technology Prathiba Sugumaran in the Platform Keynote at Snowflake AI Data Cloud Summit 2024.

In the case of Cash App, the main driver was customer behavior modeling. “Our company creates value by allowing money to flow from one customer to the next,” explained Christian Figueroa, Head of Network Science and Behavior Modeling at Cash App, in the Builder’s Keynote at Snowflake AI Data Cloud Summit 2024. “And that’s why we started working with RelationalAI as a way to look at our customer base, observe them, observe our graph and draw insights about the most important nodes in the network and how they’re connected to each other.” 

The Cash App team started by putting RelationalAI’s methods to the test. They picked two classes of algorithms that are common in network analytics: community detection algorithms that are used to segment customers into different groups, and centrality analysis that are used to identify the most important nodes in the network. These two algorithms are well-known for being resource intensive and time consuming.

“The results were pretty impressive: we saw a 10-fold reduction in both compute time and compute costs,” said Figueroa.

According to Figueroa, by getting the RelationalAI tool through Snowflake Marketplace, “We can both go far with the right tool and move at a relatively quick pace because we don’t have that upstart cost that comes with new tools.”

For RelationalAI, using the Snowflake Native Application Framework integrated with Snowpark Container Services offered three major benefits: ease of deployment and maintenance; a secured, trusted data governance foundation; and access to Snowflake Marketplace.

Without the level of built-in infrastructure provided by Snowflake Native Apps and Snowpark Container Services, onboarding customers and getting them into RelationalAI would be much more difficult, according to Macintyre. “It's very difficult to increase the utility of an organization's data when the first step is to move the data from Snowflake to another system. This really slows you down and is often simply not possible,” he says.  

“A lot of graph platforms are not scalable,” says Aref. “Sometimes that has actually turned people off of graphs …. [But] our scalability is a function of the architecture that [RelationalAI and Snowflake] both have: cloud-native with separate storage from compute.” 

The ability to execute complex AI workloads without moving data around and Snowflake’s built-in data governance foundation offers the security that companies require today. 

“Getting these workloads closer to the data makes it much easier for customers to not only use the technology, but adopt it — particularly in cases where data governance and data security is of high concern,” says Macintyre.

Aref says RelationalAI saw new customer interest and demand grow exponentially, even when their product was available only in private preview. That was, in no small part, due to Snowflake Marketplace, the central hub where hundreds of third-party providers offer data sets, applications and AI products that seamlessly integrate with the AI Data Cloud. 

RelationalAI’s Head of Engineering Reto Kramer points to the Snowflake Marketplace capabilities that offer billing options and allow the company to distribute and perform upgrades easily as a significant benefit to building on Snowflake. “It's the kind of undifferentiated development that we love for Snowflake to provide as a capability of the platform, because it’s not our core expertise,” he says.

The team also gives credit to the strong relationships Snowflake has built with its customers, seeing the benefits of that trust echo in the procurement process for RelationalAI’s own product. On a practical level, Snowflake Marketplace helps smooth out the billing process for both RelationalAI and its potential customers, who can consolidate charges into one place and have greater visibility over their aggregate spend. Taken together, these features are proving to be powerful drivers of business for RelationalAI.

Plenty of customers have already seen the value in RelationalAI’s knowledge graph coprocessor; check it out for yourself on Snowflake Marketplace.

For more about RelationalAI and to hear about the real-world impact of their technology and Snowflake Native Apps, watch the on-demand keynotes from Snowflake AI Data Cloud Summit 2024, featuring discussions with AT&T and Cash App as well as demos of RelationaAI’s tools:

• Platform keynote: Hear how AT&T uses RelationalAI to fight fraud and maximize uptime, and see a demo of how RelationalAI’s knowledge graph coprocessor app helps visualize and analyze dependencies and risks in a cell tower network.
• Builder keynote: Learn how Cash App uses RelationalAI to identify top customers and keep them engaged.