Data Analytics

Snowflake Summit 2023: How Snowflake Is Helping Mine Value from Financial Data

Financial services companies handle a high volume of sensitive data, but with the right platform and protections in place, they can leverage it to benefit themselves and their customers.

Joe Kuehne

Joe Kuehne is an Associate Editorial Director at BizTech magazine.

Financial institutions have always handled a great deal of personal and sensitive data, and the need for privacy and security can sometimes stand in the way of innovation.

However, Snowflake has been helping financial organizations securely modernize the way they manage the valuable data they possess.

As Christian Kleinerman, senior vice president of product management at Snowflake, explained during this year’s Snowflake Summit, the company introduced its Financial Services Data Cloud back in 2021. The industry-specific platform combines the core technology that Snowflake brings with data sets from the marketplace that may be useful to the financial services industry.

There are multiple use cases for financial data in Snowflake’s cloud — including fraud detection and credit underwriting — and the company isn’t trying to sell a product that performs these tasks end to end. Instead, Kleinerman said, “We want at the end of the day to be able to simplify the use cases that are coming to a given industry.”

Click the banner below to learn how a modern data platform supports smart decision-making.

Snowflake Introduces New Capabilities to Empower Financial Data

During the opening keynote of the summit, Frank Slootman, Snowflake’s chairman and CEO, and Kleinerman shared a series of new ways the company is enabling financial services companies to realize the value of their data.

Kleinerman elaborated on some of the announcements he thought would be most promising for financial institutions, highlighting the Native Apps framework in particular.

“The financials folks are at the forefront of data sharing. We see amazing success there. Many of the data providers that we have realize the need to provide not just raw data, but also an experience for business users to browse and understand the data that is included,” Kleinerman said.

DISCOVER: Find out how CDW can help your organization with data analytics solutions.

Kleinerman expressed his excitement for the potential of data and app sharing in the Snowflake Marketplace, especially light of the announcement made by Bloomberg during the summit.

Bloomberg announced in a press release “that its Data License Plus (DL+) offering now powers a Snowflake Native App that will allow mutual customers to provision their set of Bloomberg Data License subscriptions in the Snowflake Data Cloud and configure a ready-to-use environment that is already hydrated with fully modeled Bloomberg data and multivendor ESG content — all within minutes.”

He also spoke highly about the potential for Snowflake Container Services within the financial sector. “I think it significantly changes the art of what's possible with Snowflake. And I think it's equally interesting to partners, building solutions and distributing tools — not just partners, but customers as well.”

He offered the example of Goldman Sachs, which recently integrated their open-source platform Legend with Snowflake and AWS. That Goldman Sachs data, according to Kleinerman, is now “included in a container that will be made available to other financial services customers as part of this platform.”

Many of the data providers we have realize the need to provide not just raw data, but an experience for business users.”

Christian Kleinerman Senior Vice President of Product Management, Snowflake

Financial Data Demands Careful Attention to Privacy

Kleinerman also shared Snowflake’s awareness and understanding of the sensitive nature of data in the financial sector.

“I made an effort at the keynote to cover the comprehensiveness of our platform. It starts with being able to identify what sensitive data you may have in a data set. We also have a classification that helps you identify what we call quasi-sensitive identifiers,” he explained.

“The classification we have helps you look at not only columns that are identifying, but also combinations of columns that can be used for identification.”

The potential manipulation of identifying data led Snowflake’s developers to provide their users the ability to mask, partially mask or conditionally mask data, depending on who’s querying it.

“For some really strict scenarios, masking may not be sufficient, or you may end up losing too much analytical value. The mathematically provable answer to preserve privacy but retain analytical value is differential privacy,” Kleinerman said. That’s why Snowflake acquired LeapYear earlier this year, to add differential privacy to its platform.

In addition, Kleinerman said the last element of the company’s suite of privacy capabilities is to enable users to share data functions instead of sharing the data itself. “And through the functions, you can control exactly what type of questions can be asked of your data. That sharing of functions is the foundation of what we call Clean Rooms. And we've been at the forefront of providing secure and private multiparty collaboration.”

“It is paramount that you provide that human oversight. Because there's also risk with these new technologies, right?”

Torsten Grabs Senior Director of Product Management, Snowflake

AI Offers New Possibilities for Financial Institutions

In a related conversation at the summit, Torsten Grabs, senior director of product management at Snowflake, offered some use cases for how the financial services industry could employ Snowflake’s new artificial intelligence and machine learning capabilities.

As part of the event’s opening keynote, Slootman spoke about Snowflake’s new Document AI capabilities, and Grabs noted that the technology is already being used successfully for accelerating and automating mortgage approval processes.

Performing mortgage approvals requires the acquisition of multiple documents, such as pay stubs and tax records, that have some reasonable structure to them.

Grabs said that Document AI will allow financial institutions to easily understand these documents via automated processes and “then extract the salient information from these documents so you can understand the financial situation of the customer.”

Once a financial institution has developed its own language model for the process, it can be used to score the customer and then suggest whether to approve or decline them for a particular mortgage.

FIND OUT: Learn how businesses should be thinking about AI with digital transformation.

“What's great about what we announced today is, as you find those cases where you don't agree with what the system suggests, now you have the opportunity to actually go back and provide feedback, which then leads to fine-tuning of this large language model that we're running for you. So increasingly, over time, the model will get better and better and better.”

Grabs did offer a note of caution about the potential elimination of human involvement in the process. “It is paramount that you provide that human oversight and you keep the human in the loop. Because there's also risk with these new technologies, right?” he said.

“You need that oversight in order to make sure that your organization can fully stand behind what's coming out of this service. I think the way you should think about these productivity enhancements is as accelerators for the work that the humans are doing — as assistants, as companions that help the employees in an organization. But they're not completely automating away those people.”

Keep this page bookmarked for articles and videos from the event, follow us on Twitter @BizTechMagazine and join the event conversation at #SnowflakeSummit.