In this interview from Snowflake Summit 2026, Sherene Hilal, chief ad product and technology officer at SiriusXM, joins Dennis Buchheim, global head of marketing technology, media and entertainment at Snowflake, to talk with theCUBE's Dave Vellante and Rebecca Knight about how data-driven personalization and open ecosystem partnerships are redefining audio media at scale. Hilal explains how audio — present in the foreground and background of roughly four hours of daily listening — has long lacked the structured data needed for real-time targeting and discovery. She outlines how SiriusXM is addressing that through deep data taxonomy: analyzing song lyrics, listener context, time of day and podcast transcripts to deliver tailored experiences across every environment. Buchheim connects this to Snowflake's newly announced Cortex Sense, a semantic intelligence layer that defines meaning across an organization's data corpus and builds toward full Customer 360 personalization and ad targeting.

The conversation also explores SiriusXM's push to build an open ecosystem — including a first-of-its-kind ad representation deal with YouTube that goes beyond inventory sales to co-developing audio infrastructure at scale. Hilal describes how multimodal media, where listeners lean in for video and lean back into audio, is being measured and monetized through cross-platform partnerships with Google and standards bodies like the Interactive Advertising Bureau. The discussion shifts to AI's most immediate application inside SiriusXM: speed. By standardizing disparate supply types across satellite, streaming and digital platforms, AI is eliminating the manual processes that once slowed content distribution. Buchheim highlights how Snowflake Intelligence and CoWork are democratizing data access for business users — compressing what once took a week of analyst work into real-time agentic conversations. From replacing keyword search with data-driven discovery to evolving product planning into outcome-driven KPI frameworks, both speakers outline how media and enterprise organizations can move from AI experimentation to governed, production-ready intelligence.

In this interview from Snowflake Summit 2026, Anahita Tafvizi, chief data and AI officer of Snowflake, joins Durgesh Das, vice president of data, analytics and governance at ICE and NYSE, to talk with theCUBE's Dave Vellante and Rebecca Knight about moving enterprise AI from fragmented experimentation to governed, production-ready agentic intelligence. Tafvizi details Snowflake's 18-month journey to becoming a fully AI-native organization, anchored by a go-to-market assistant now fielding more than 40,000 questions per week across the sales team. She walks through CoCo and CoWork — both announced at the summit — explaining how each closes the gap between governed data and AI-driven action. Das describes ICE's live deployments, including a customer 360 skill that collapsed five Tableau dashboards into a single conversational interface and cut pre-meeting prep from a full day to minutes.

The conversation also explores how enterprises can avoid rebuilding islands of automation as AI access expands. Tafvizi underscores the importance of semantic layers, role-based access control and trusted data foundations as the guardrails that let teams experiment freely without generating inaccurate or ungoverned results. Das details ICE's deliberate approach to token governance — tying consumption directly to ROI by tracking deliverables rather than raw usage — and notes that cost intelligence is built directly into CoCo as a skill. From automating overnight data pipeline repairs with AI agents that diagnose issues, submit pull requests and queue approvals by morning to reframing governance as an accelerator rather than a constraint, both leaders chart a practical path from data to insights to action for enterprises ready to move beyond the dashboard.

In this interview from Snowflake Summit 2026, Bala Kasiviswanathan, vice president of developer and AI experiences at Snowflake, joins Caitlin Halferty, head of data and analytics at Thomson Reuters, and Laura Safdie, head of legal innovation at Thomson Reuters, to talk with theCUBE's Dave Vellante about how a governed data foundation makes the leap from AI experimentation to enterprise-ready agentic intelligence possible. Kasiviswanathan anchors the discussion around two announcements: CoWork, which gives business users governed data access through natural language, and CoCo, a builder-focused development tool already used by over 1,500 Thomson Reuters employees. Halferty frames Thomson Reuters' approach through the lens of "fiduciary grade" AI — four pillars of authoritative content, data security, expert validation and transparent outputs — explaining why professionals who carry personal liability for their decisions demand nothing less.

The conversation also explores how Safdie and the CoCounsel team evolved their legal AI assistant from RAG-based retrieval through an agentic harness natively built into Westlaw and Practical Law, enabling lawyers to delegate complex litigation planning while staying accountable at every step. Halferty details how consolidating 23 disparate customer data sources into a unified semantic intelligence layer eliminated data conflicts and gave sellers actionable insight on retention signals spanning product usage, billing and customer sentiment. The panel also touches on Thomson Reuters' acquisition strategy — including the Casetext deal that brought Safdie into the company — as a deliberate path to injecting AI-native culture into a 170-year-old institution. From Snowflake's newly announced Cortex Sense runtime to the principle that governance is an accelerant rather than a constraint, the discussion provides a practical blueprint for established enterprises racing to become AI-native without compromising the trust their most demanding customers require.

In this interview from Snowflake Summit 2026, Vivek Raghunathan, senior vice president of engineering and support at Snowflake, joins Matt Luizzi, vice president of analytics at WHOOP, to talk with theCUBE's Dave Vellante and Rebecca Knight about the path from fragmented data and AI initiatives to governed, production-ready agentic systems. Luizzi details how WHOOP manages over three petabytes of sensor data — growing at 20 terabytes per day — and relies on Snowflake as the foundation for business analytics and accelerating insight delivery. He explains why interoperability drives the architecture, pointing to WHOOP's adoption of Iceberg and Polaris as guardrails for evolving without lock-in. Raghunathan frames Snowflake's broader thesis: rather than fragmenting AI across disparate tools, the platform extends to cover both business users through Snowflake CoWork and developers through Snowflake CoCo.

The conversation also explores Snowflake's four-level agentic maturity model — from extending CoCo with custom skills to event-driven automation, workflow orchestration and continuously learning systems that capture tribal knowledge across the enterprise. Raghunathan highlights Cortex Sense, born from a December hackathon, as Snowflake's answer to the intelligence layer enterprises need to connect governed data to coding agents, and points to the Natoma acquisition as its solution for managing the proliferating maze of MCP connections. Luizzi offers a ground-level view from WHOOP, describing how agentic tools are already compressing role boundaries — blurring the lines between product managers, analysts and engineers — while keeping humans focused on strategic judgment. Raghunathan adds a first-principles challenge to the org-chart conversation, arguing that the more pressing question is identifying which activities can be fully autonomous, which require human judgment and which remain fundamentally creative. From Cortex Sense as a new enterprise intelligence layer to WHOOP's emerging agent hierarchy mirroring a corporate org chart, the discussion provides a concrete roadmap for enterprises navigating the shift from AI experimentation to governed, self-improving agentic systems.

In this interview from Snowflake Summit 2026, Dwarak Rajagopal, vice president of AI engineering and research at Snowflake, joins Spiros Xanthos, founder and chief executive officer of Resolve AI, to talk with theCUBE's Dave Vellante and Rebecca Knight about how governed model training and purpose-built AI agents are closing the gap between AI experimentation and production-ready reliability. Xanthos describes Resolve AI as an SRE agent — an autonomous system that takes over on-call responsibilities by picking up production incidents, diagnosing root causes and proposing fixes before human engineers need to intervene. Rajagopal explains how Snowflake's Cortex training platform allows companies like Resolve AI to train custom models directly where their enterprise data already lives, eliminating the need to manage GPU sourcing, distributed systems or infrastructure overhead. The partnership enables Resolve to focus on what differentiates it — domain-specific reasoning in production environments — rather than rebuilding an abstraction layer that Snowflake already provides.

The conversation also explores the active debate between frontier models and specialized alternatives. Xanthos argues that production software is a uniquely hard domain: data volumes exceed context windows, reasoning chains are long and noisy, and frontier labs lack the deep operational expertise needed to build truly effective domain-specific solutions. Both guests converge on a hybrid view — frontier models handling general tasks while custom, post-trained models deliver better performance-to-price ratios for narrow, well-defined work. Rajagopal details how Snowflake's ZoRRo (Zero Redundancy Rollouts) technique improves reinforcement learning efficiency and enables multi-tenant GPU utilization, directly addressing the supply constraints that continue to pressure enterprise AI deployments. Xanthos also reveals Resolve AI's outcome-based pricing model, in which customers purchase credits consumed by agent work rather than committing to software subscriptions — a structure that only charges when agents deliver results. From redefining on-call operations to converging training and inference into a unified platform strategy, the discussion maps a practical path for how AI-native companies and enterprise platforms can build specialized, production-grade intelligence together.

In this interview from Snowflake Summit 2026, Dayne Turbitt, senior vice president of EMEA at Snowflake, joins Emmanuel Frenehard, chief digital officer of Sanofi, to talk with theCUBE's Dave Vellante and Rebecca Knight about how enterprises are moving from AI experimentation to governed, production-ready agentic intelligence. Frenehard frames Sanofi's AI-first strategy around a single imperative: compressing a 10-to-12-year drug development cycle, where 90% of drugs still fail at phase three, down to five or six years through simulation and data-driven precision. Turbitt connects this ambition to a broader market pattern — organizations that have done the hard work of unifying data under governance are now positioned to unleash agents that reimagine entire workflows, treating AI as a coworker rather than a productivity tool.

The conversation also explores how Sanofi anchors its agentic approach in a single, governed data environment on Snowflake — which Frenehard describes not as a data platform but as an operating system — eliminating intermediary systems so that AI workflows run directly on trusted data. Turbitt details how Snowflake is applying the same principle internally through its Raven product, enabling sales teams to arrive at client conversations more informed through AI-driven research and next-best-action recommendations drawn from patterns across 12,000 customers. Frenehard also highlights the cultural dimension, noting that 44,000 of Sanofi's 75,000 employees voluntarily participated in hands-on AI workshops — a signal of the organizational readiness required to move from optimizing existing processes to reinventing them entirely. From simulating molecule behavior and virtual clinical trials to building joint IT-business teams of "expert generalists," both leaders outline a practical roadmap for enterprises ready to turn governed data into a foundation for transformation at scale.

In this interview from Snowflake Summit 2026, Sanjeev Mohan, founder and owner of SanjMo, joins Bob O'Donnell, founder and chief analyst of TECHnalysis Research, to talk with theCUBE's Dave Vellante about the shift from data-centric experimentation to production-ready agentic AI. The analysts reflect on a notable theme reversal at this year's Summit: where Apache Iceberg and open table formats drove prior conversations, agentic AI is now the central story. O'Donnell highlights how GenAI has finally delivered on big data's long-deferred promise, enabling organizations to extract actionable intelligence from data that was previously impossible to query at scale. Mohan adds a structural counterpoint — arguing that as federation matures and egress costs disappear, the competitive moat has shifted from data storage to the metadata and context layer above it, where business semantics and governance are defined.

The conversation also explores what it takes to give agents the grounding they need to operate across fragmented, multi-format data environments. Mohan breaks down the difference between brute-force retrieval and genuine contextual understanding, explaining why semantics, taxonomy, ontology and knowledge graphs must be properly layered before agents can reliably handle complex business queries. Both analysts weigh in on the limits of AI judgment — debating whether true autonomous decision-making requires something close to enterprise AGI — while agreeing that platforms like Snowflake's CoWork, which learns from human reasoning traces over time, represent a practical path forward. O'Donnell brings cross-event perspective from Microsoft Build, noting parallel themes around developer tooling, specialized task-optimized models and the imperative to connect custom organizational data with frontier model capability. From Snowflake's deliberate pivot away from its own foundation models to the emergence of Natoma as an MCP-based tool for linking specialized models, the panel outlines a pragmatic, full-stack roadmap for turning fragmented AI initiatives into governed, scalable agentic systems.

In this interview from Snowflake Summit 2026, Baris Gultekin, vice president of product and AI at Snowflake, joins Nicolas Doyen, vice president of product and data strategy at Viral Nation, to talk with theCUBE's Dave Vellante and Rebecca Knight about how governed agentic AI is moving enterprises from passive analytics to intelligent systems that coordinate and take action. Gultekin walks through the three pillars of the Snowflake CoWork announcement: a personal work agent with governed connectors to email, calendar and Slack, enabled by the Natoma acquisition; an autonomous scheduling layer that handles tasks, meeting briefs and prioritization; and AI-first interactive dashboards that business users can interrogate conversationally. He also introduces Cortex Sense, a new context runtime that lets builders deploy production-ready agents without constructing semantic models from scratch.

The conversation also explores how Viral Nation is putting these capabilities to work at the intersection of social media, brand safety and data strategy. Doyen describes how the company deployed Snowflake Intelligence to analyze real-time social sentiment around major live events — parsing brand performance at a scale and speed that traditional BI tools cannot match. The discussion broadens to the agency of the future, where Doyen sees AI accelerating ideation, testing and coordination without displacing the human creativity and relationship nuance at the core of influencer marketing. Gultekin reinforces that governance and security remain foundational as agents take on greater autonomy, noting that agents are only as effective as the data they are given access to. From validating thousands of influencer activations each month to navigating the broader cultural debate around AI and creative work, the conversation maps a practical path toward AI-powered operations that are both scalable and human-centered.

In this interview from Snowflake Summit 2026, Gowtham Gundu, chief AI officer at Fetch Rewards, joins Paul Winsor, head of retail EMEA at Snowflake, to talk with theCUBE's Dave Vellante and Rebecca Knight about how governed data infrastructure and agentic AI are reshaping the future of retail commerce. Gundu, who spent nearly two decades leading AI at Google before joining Fetch Rewards, details what drew him to a platform processing 13 million receipts daily — a transactional data footprint rivaled only by Walmart and Amazon. He explains how Fetch is using Snowflake as a foundational data layer to power deep personalization, expand from enterprise into mid-market and SMB and deploy AI agents that autonomously resolve customer support issues in real time.

The conversation also explores how Winsor views 2026 as a decisive year for agentic commerce — a shift where personal AI agents may soon shop on behalf of consumers, making clean, fast and queryable data a competitive necessity for every retailer. He references a striking insight from NRF: if an agent cannot read a product's data within milliseconds, that product effectively ceases to exist. Gundu reinforces this by describing Fetch's work to surface a natural language interface over its data warehouse, allowing brand partners to query basket size trends and buyout rates on demand. From fixing the broken value exchange between CPG brands and consumers to leveraging data clean rooms for privacy-safe multi-partner sharing, both guests make clear that the minimum bar for the agentic era is data readiness — and the window to build it is now.

In this interview from Snowflake Summit 2026, Prasanna Krishnan, head of Horizon Catalog and collaboration at Snowflake, joins Christina Egea, senior vice president of product and enterprise data at Capital One, to talk with theCUBE's Dave Vellante and Rebecca Knight about how governed data sharing and agentic AI are reshaping enterprise intelligence. Krishnan details a wave of Horizon Catalog innovations designed to connect enterprise data and metadata wherever it lives — without copying or moving it — and automate the toil of governance through intent-driven policies. Egea highlights how Capital One has leveraged Snowflake's sharing capabilities to achieve granular visibility into data flows while meeting its regulatory obligations, describing it as a powerful unlock for democratizing data access across the entire organization.

The conversation also explores the security implications of agentic AI, where existing enterprise ecosystems were simply not built for agent-level identity and access management. Krishnan outlines new Horizon capabilities including agent identity, AI security posture management and protections against prompt injection — enabling organizations to set, audit and enforce policies that distinguish agent actions from human actions. Egea adds that the same data can be perfectly appropriate for one use case while off-limits for another, making context-aware, policy-driven governance essential as agents begin acting on behalf of users. The discussion also touches on Capital One's evolution from Snowflake customer to platform builder, having developed internal data management tooling it now offers externally as a financial services provider. From navigating the governance of MCP endpoints to converging data, AI and business decisions in real time, both guests outline why the next phase of enterprise intelligence demands a fundamentally reimagined approach to how data and agents interact.