Agentic AI is rewriting the rules of enterprise data engineering, turning passive infrastructure into intelligent systems that act, adapt and automate at scale.

Gone are the days when platforms simply stored and queried data. Today’s data stacks are being rearchitected to support agentic workflows — systems where metadata, governance and orchestration layers work in concert to guide machine reasoning. Snowflake Inc.’s embrace of open formats such as Apache Iceberg signals a broader shift: Enterprises are building environments where intelligence is embedded into the fabric of data itself. This transformation is accelerating innovation across cybersecurity, finance, advertising and beyond, as organizations seek not just insight, but action, according to Dave Vellante (pictured, right), co-founder and chief analyst at theCUBE Research.

TheCUBE’s Dave Vellante discusses agentic AI.

“The value is moving up the stack, and as they lean into open table formats like Iceberg, there’s more of the proprietary Horizon functionality trickling into the open world,” Vellante said. “That puts greater pressure on Snowflake to have the absolute best engine … and also to be the best at managing all those open table formats.”

Vellante, along with theCUBE’s George Gilbert (left) and Rebecca Knight (center) spoke with experts from Snowflake and other companies at Snowflake Summit, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how agentic AI is transforming enterprise data engineering by embedding intelligence directly into data platforms, enabling autonomous decision-making and real-time action across industries such as cybersecurity, finance and advertising. (* Disclosure below.)

Here’s theCUBE’s complete day one keynote analysis:

Insight #1: Metadata standardization is reshaping agentic AI platforms.

In today’s enterprise data engineering landscape, platforms are evolving rapidly to support AI agents that rely on standardized metadata and flexible compute strategies. This pivot is elevating metadata from a passive reference to an active control layer for insight delivery.

“Their engine can manage read-write access, they can do data engineering, analytics, data science, even on external Iceberg tables,” Vellante said. “That’s a huge step toward making the open experience feel like the Snowflake experience.”

SanjMo’s Sanjeev Mohan discusses putting in place a common metadata standard.

The convergence of Horizon and Polaris catalogs is central to this evolution. By synchronizing permissions and lineage across native and open table formats, Snowflake is enabling its engine to treat external Iceberg tables with nearly the same fidelity as native ones. This enhances not only performance and governance but also trust, an essential foundation for agentic workflows that must reason, decide and act autonomously, according to Gilbert.

“The metric and dimension layer is the human language, and that gets translated into SQL,” he said. “They’ve put a lot of effort into this because when you want to talk to your data, it’s a feature not to have to use dashboards or SQL.”

Here’s theCUBE’s complete insights:

These architectural changes are not just cosmetic. They mark a turning point where platforms are seen not as passive repositories, but as intelligent, agentic participants in enterprise decision-making, according to Sanjeev Mohan, principal at SanjMo, during an analyst segment at Snowflake Summit.

“Every specialized super micro segment of the landscape or the pipeline had 20 different players,” he said. “We just talked about [how] there’s no common metadata standard, which means that you’re constantly integrating and there’s so much overhead. The whole idea why I think this would be successful is because customers are saying, ‘We want minimal overhead.’ What AWS would call undifferentiated heavy lifting.”

Here’s theCUBE’s complete interview with Sanjeev Mohan:

Insight #2: Industry partnerships are extending AI engineering capabilities.

As agentic AI platforms evolve, industry leaders are doubling down on collaborative development models. Partnerships are emerging as the fastest route to integrate domain-specific AI capabilities, especially in infrastructure, financial services and unstructured data processing.

Meta’s Amit Sangani talks to theCUBE about his company’s work with Cortex AI.

“Snowflake has been our partner ever since day zero,” said Amit Sangani, senior director of AI platform engineering at Meta Platforms Inc. “We have partnered with Snowflake on Llama 2, Llama 3. Now just we recently launched Llama 4 models and Snowflake has been our partner. We are working very closely; it’s integrated in their Cortex AI platform where we are basically making sure that the models are working very well across the agentic systems.”

This collaboration enables Snowflake users to interact with leading open-source LLMs through its Cortex AI platform, blending ease of deployment with model transparency. Meta gains an enterprise delivery channel while Snowflake expands its generative AI interface.

“Our customers use Llama via [the] Cortex platform,” said Dwarak Rajagopal, VP of AI engineering and research at Snowflake. “They’ve been using Llama models via the Cortex to actually power personalized recommendation, travel recommendations for their users and have seen big both positive customer engagement, but also big business impact as well.”

Here’s theCUBE’s complete interview with Amit Sangani and Dwarak Rajagopal:

In the financial sector, Deloitte Consulting LLP is bridging compliance and AI by using governed platforms to fast-track adoption for highly regulated clients. Their goal: Reduce the typical friction that arises from risk, control and data complexity.

“The challenge is not having data,” said Shailender Sidhu, AI and data principal, financial services industry, at Deloitte. “The challenge is how do you stitch the data together which creates meaningful insights?”

Partners are also helping address the challenges posed by unstructured data, a critical piece for scalable AI. Integrating structured and unstructured data into a single ecosystem is essential for AI’s next leap, according to Box CTO Ben Kus in an interview with theCUBE.

“With generative AI, it unlocks a lot of the idea that generative AI was born on unstructured data,” he explained. “It can not only help you understand it, but then also help you do things like pull data, extract data from it, answer questions about it and so on.”

Here’s theCUBE’s complete interview with Ben Kus and Chris Child, vice president of product, data engineering at Snowflake:

Insight #3: Real-world use cases are showing the impact of agentic AI.

Agentic AI is not a future ideal; it’s already reshaping enterprise workflows across sectors such as finance, cybersecurity and internal analytics. The common thread: delivering decision-ready insights, faster and more responsibly.

Rakuten Rewards’ Adam DeMonaco and Snowflake’s Brad Jones talk about agentic AI with theCUBE.

“Our big focus has really been on centralizing our data and then making the best use of our data from an analytical perspective,” said Adam DeMonaco, chief information security officer of Ebates Performance Marketing Inc., d/b/a Rakuten Rewards, in an interview with theCUBE. “We’ve been able to move from a distributed model to more of a centralized model and then centralize security controls based on classification through discovery and been able to really centralize that platform quite a bit.”

Cybersecurity teams are using agentic AI to orchestrate signal triage, reduce alert fatigue and deliver contextual insights to analysts. This reduces manual overhead while improving response accuracy.

“We think we’re best positioned to do that by having a uniform governance and control structure around all of the data from ingestion through business insights,” said Brad Jones, chief information security officer and vice president of information security at Snowflake. “The challenge is when you have third-party tool sprawl, you’re trying to take that same governance and control structure and map it to disparate platforms. By having a uniform control structure, governance structure that goes multicloud, multi-CSP with Snowflake, you get that uniformity. And then you can allow users to innovate with AI because you know it’s working within your security boundary.”

Here’s theCUBE’s complete interview with Brad Jones and Adam DeMonaco:

On the financial side, S&P Global Inc. is leveraging Snowflake’s AI tooling to analyze 192,000 earnings calls over 17 years. The result: a nuanced, data-driven understanding of management behavior and market signals.

“When an executive is answering a question, do they remain on topic to the question asked or do they go off topic,” asked Liam Hynes, head of new product development at S&P Global. “What we found was that firms with executives who stay on topic to the questions asked, outperform their off-topic peers.”

Snowflake embeds AI directly into every team’s workflow rather than isolating it, focusing on semantic modeling, verified queries and data trust. Internally, the company uses its own AI assistant to ensure 100% accuracy for decision-making, emphasizing that trust, governance and quality must underpin AI-powered tools, according to Anahita Tafvizi, chief data analytics officer at Snowflake. Snowflake is embedding agentic capabilities directly into operational workflows.

“AI has to be everyone’s job, not just one team’s job,” she said. “We kept it within every team, embedded in every team, and we want everyone to think about how to future-proof the technology, how to future-proof the solutions that we are building and basically look at AI as part of their early job.”

Here’s theCUBE’s complete interview with Anahita Tafvizi:

To watch more of theCUBE’s coverage of Snowflake Summit, here’s our complete event video playlist:

(* Disclosure: TheCUBE is a paid media partner for Snowflake Summit. Neither Snowflake Inc., the primary sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE