Data Analytics

Snowflake Summit 2025: How To Make Data and Applications More Trusted and Connected With AI

Snowflake announced several new enhancements, from semantic models to masking data to ensure AI is accurate at scale.

Twitter

Lily is a Senior Editor at BizTech Magazine. She follows tech trends, thought leadership and data analytics.

“This looks like a rock concert, but for data people,” said Conviction CEO Sarah Guo at the opening keynote of Snowflake Summit 2025, hosted in San Francisco. A crowd of 20,000 erupted in cheers as speakers unveiled bold visions for an agentic future powered by terabytes or even petabytes of data.

But all of that data comes with great responsibility. Not only does it need to be clean, trusted and scalable but it also must deliver on its promise. “That’s why data readiness isn’t just a technical project,” Snowflake CEO Sridhar Ramaswamy told Forbes. “It’s the foundation of whether your AI investments will even work.

Experts at the summit said that trust starts with building transparent data pipelines and governance frameworks that access controls and compliance checks at every level. Only then can teams be sure their large language model outputs will be effective.

Here are a few ways enterprises can ensure that artificial intelligence is accurate.

Click the banner for related insights on how AI and data analytics can fuel your business.

Build Trust Through Transparent AI Use and Governance

One way to build trustworthy AI is to make sure the data is being governed from the start. “If you ask our customers today, they are all very concerned about security and governance. And where does security governance start? It really starts with the data that feeds into these LLM models,” said Artin Avanes, head of core data platforms at Snowflake.

According to Christian Kleinerman, senior vice president of product at Snowflake, that just became easier now that Snowflake’s Copilot AI assistant is embedded into Horizon’s governance platform.

“What we have done is look at the entire lifecycle of governance,” Avanes said. Users can now discover, label and analyze structured and unstructured data, with built-in tools to mask personally identifiable information data for stronger privacy and compliance. This ensure that large language models are only pulling from relevant data and avoiding proprietary sets.

“This is the next frontier,” Avanes said, because it will allow highly regulated industries such as finance to customize their compliance checks in the data stream.

If you ask our customers today, they are all very concerned about security and governance. And where does security governance start? It really starts with the data that feeds into these LLM models.”

Artin Avanes Head of Core Data Platforms, Snowflake

Lynn Martin, president of NYSE Group, said that generative AI has helped ensure that model outputs align with regulatory standards and that sensitive market data is shared with strict governance policies. It’s also helped her teams maintain confidence that what they share with stakeholders and investors is accurate.

Trust is paramount. “How do you verify the results? How do you give citations so that you know these agents are not making up data?” Kleinerman said. Snowflake can now provide that record by turning any text into SQL.

“Say you ask a question in natural language, you can see the text turn into SQL, and you can see where it’s pulling this data from and inspect it,” said Prasanna Krishnan, head of collaboration and Horizon at Snowflake.

The “creation of semantic models for structured, tabular data” has also helped users ask questions and understand how the model fetches these results, Krishnan added.

A CIO can see what ways people are running the data, which role-based accesses are running these queries, and that becomes really important from a compliance perspective.”

Prasanna Krishnan Head of Collaboration and Horizon at Snowflake

Ensure AI Delivers ROI With Real-Time, Relevant Insights

If users see small and iterative ROI on AI, that too instills trust. “It should be a series of little projects that show value every step of the way,” Ramaswamy told Forbes.

PayPal uses Snowflake Copilot and Snowflake Cortex to detect anomalies in real time across billions of transactions. According to Avi Reddy, director of post processing for PayPal, the company has been able to catch fraud faster, reduce false positives and deliver clearer insights to security teams.

Enterprises can also leverage the newly announced Snowflake Intelligence “to harness real-time insights across news, research, and publications to enrich their AI outputs,” according to a company press release.

If teams are still skeptical, they can run their data through Snowflake’s Trust Center, where third-party security and observability scanners can flag abnormalities.

EXPLORE: AI in the workplace is empowering people to do more with less.

Allow AI and Computing Power To Solve Real-Life Business Challenges

Another powerful way to build trust in AI is by aligning it with the real challenges employees face. Doing so not only resolves pain points but also makes the technology’s value more tangible to teams.

“So, if you think about where are the big problems and you layer that alongside what your principles are, your use cases will follow,” Martin said.

The hospitality chain Marriott, which has over 228 million loyalty program members, couldn’t get a sense of who these travelers were as individuals. What are their travel patterns or their hotel preferences?

UP NEXT: What are the benefits of unsupervised machine learning and clustering?

“We have billions of data records flowing in real time,” said Julia Morrison, vice president of data and personalization at Marriott. Her team is using AI to personalize guest experiences, from recommending rooms to offering services based on traveler behavior.

But guests expect a degree of anonymity at hotels, and that’s where strict data privacy controls come into play. Marriott relies on Snowflake’s data masking and secure sharing.

As teams work with data at this granular level, IT leaders can gain a macro understanding of how the data is being used across the organization. “A CIO can see what ways people are running the data, which role-based accesses are running these queries, and that becomes really important from a compliance perspective,” Krishnan said.

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