From systems of intelligence to systems of action
Year: 2026 · ▶ Watch on YouTube
Stephanie Wong (Global Lead, Developer Programs) · Yasmeen Ahmad (Managing Director, Data Cloud)
Segments (13)
- 00:00:00 · Introduction — Stephanie Wong
- Stephanie Wong introduces Yasmeen Ahmad to discuss the latest in Google’s Data Cloud.
- 00:00:18 · From System of Intelligence to System of Action — Yasmeen Ahmad
- The discussion begins on how the role of data is evolving from providing intelligence to enabling direct action through agentic systems.
- 01:12 · The Importance of Data Strategy and Context — Yasmeen Ahmad
- Ahmad explains that a successful agentic AI strategy requires not just clean data, but also rich, machine-readable business context that was previously ‘invisible work’.
- 03:50 · The Role of Semantic Understanding — Stephanie Wong
- Wong asks if the transition is enabled by AI agents’ new ability to have a semantic, inferred understanding of data.
- 05:52 · Gemini Enterprise and Existing Data Assets — Stephanie Wong
- The conversation shifts to how Gemini Enterprise acts as a new front door, enabling organizations to activate their BigQuery and Looker assets.
- 06:24 · Integrating Data Cloud with Gemini Enterprise — Yasmeen Ahmad
- Ahmad details how integrations allow business users to chat with their data through Gemini Enterprise without needing to understand the underlying data platforms.
- 07:04 · Deep Research Agent Integration — Yasmeen Ahmad
- Ahmad describes how the Deep Research Agent is connected to the Knowledge Catalog to reason over both enterprise and web data for richer insights.
- 08:23 · The Power of AI Agents and the Data Stack — Stephanie Wong
- Wong highlights that AI agents represent a powerful, flexible layer in the stack for executing actions like function calling and deep research.
- 08:56 · Addressing Scattered Data with Cross-Cloud Lakehouse — Stephanie Wong
- The challenge of scattered data is raised, questioning how Google’s Cross-Cloud Lakehouse and support for open standards like Apache Iceberg help.
- 09:53 · Cost and Performance at Agent Scale — Stephanie Wong
- Wong asks how Google’s AI-optimized infrastructure and serverless approach are helping customers scale their AI ambitions efficiently.
- 10:06 · Optimizing the Full Stack for Efficiency — Yasmeen Ahmad
- Ahmad explains that Google optimizes every layer of the stack, from silicon to data engines, to handle the increased load from agent swarms efficiently.
- 19:02 · The Future of the Agentic Era — Stephanie Wong
- Wong asks what Ahmad is most excited about in the new agentic era for the Data Cloud.
- 19:10 · Conclusion: Seeing Systems of Action in Practice — Yasmeen Ahmad
- Ahmad expresses excitement about seeing customers already implementing swarms of agents to drive true action and achieve previously impossible ROI.
Products Announced (6)
- 05:52 ·
Gemini Enterprise(Mentioned)- Acts as a ‘new front door’ for data · Integrates with BigQuery and Looker assets · Enables conversational interaction with business data
- 10:18 ·
BigQuery Spark (with Lightning Engine)(New Feature (Lightning Engine))- 5x faster than plain vanilla Apache Spark · 2x better price-performance than market alternatives · Managed service for Apache Spark
- 10:18 ·
Data Agent Kit(Announced)- Provides plugins, extensions, tools, and skills for AI agents · Enables agents to natively understand Google’s Data Cloud · Connects agents to take action across systems
- 11:03 ·
Cross-Cloud Lakehouse(Mentioned)- Provides a single universal plane to see all data · Supports open standards like Apache Iceberg · Connects to data where it lives without requiring movement
- 12:53 ·
Knowledge Catalog(Mentioned)- Creates inferred schemas and descriptions for unstructured data · Connects with Deep Research Agent · Provides a universal context engine for agents
- 13:04 ·
Deep Research Agent(New Integration)- Connected to the Knowledge Catalog · Reasons over enterprise data, web data, and documents · Provides deep, rich, and holistic answers
Competitor Mentions / Comparisons (3)
- 11:18 · vs AWS — Mentioned as a multi-cloud environment where customers run SaaS applications or store data (S3 Glue) that can be accessed via Google’s cross-cloud capabilities.
- 12:04 · vs Databricks — Mentioned as a data platform whose Unity Catalog can be reached into by Google’s cross-cloud solutions.
- 12:06 · vs Snowflake — Mentioned as a data platform whose Polaris catalog can be reached into by Google’s cross-cloud solutions.
Benchmarks Shown (5)
- 15:26 ·
BigQuery Performance Improvement: 35% improvement- Over the last year
- 15:35 ·
BigQuery Cost Reduction: 40% reduction- Over the last year
- 16:25 ·
BigQuery AI Inferencing Efficiency: 230x- When running AI inferencing over BigQuery data
- 15:48 ·
Apache Spark with Lightning Engine Performance: 5x faster- Plain vanilla Apache Spark
- 15:52 ·
Apache Spark with Lightning Engine Price-Performance: 2x better- Market proprietary alternatives
Notable Quotes (6)
- 00:21 — Stephanie Wong:
The system of intelligence really is changing and evolving into a system of action.
- 01:12 — Yasmeen Ahmad:
What we see with generative AI, and in particular now these agentic systems, is driving action is much easier.
- 03:20 — Yasmeen Ahmad:
That context was never built into data platforms. That context was what I call invisible work that was outside the data platform in the human mind.
- 08:58 — Stephanie Wong:
But the challenge though is that data still can be scattered across many places and environments.
- 14:13 — Yasmeen Ahmad:
As these agents come online, they are hungry. And it’s not just single agents, it’s swarms of agents that we are seeing.
- 17:11 — Yasmeen Ahmad:
Only Google is working on infrastructure, the model innovation, the data innovation, all together.
Visual Signals
On-screen (4)
- 00:00:00 ·
Google Cloud Next '26 logo- Brands the event and the year.
- 00:04:08 ·
Lower third: Stephanie Wong, Global Lead, Developer Programs, Google Cloud- Identifies the host and her role.
- 00:42:08 ·
Lower third: Yasmeen Ahmad, Managing Director, Data Cloud, Google Cloud - 20:21:23 ·
Cloud Next '26 logo on a white background- Outro branding for the video segment.
Stage (2)
- 00:00:00 · The video opens on Stephanie Wong in a studio setting, sitting at a desk with a microphone and headphones.
- 00:11:11 · The camera angle switches to a wide shot showing both Stephanie Wong and Yasmeen Ahmad seated across from each other at a desk labeled ‘Google Cloud Next’.
Key Topics
Agentic AI · Data Cloud · System of Action · Generative AI · Data Strategy · Multi-cloud · Cross-cloud Lakehouse · BigQuery · Gemini Enterprise · Apache Iceberg · Data Governance · AI Agents · Knowledge Catalog · Cost Optimization · AI Infrastructure
Takeaways
- The paradigm is shifting from ‘systems of intelligence’ (providing insights) to ‘systems of action’ (AI agents executing tasks), which is enabled by generative AI.
- A successful agentic AI strategy requires a robust data foundation that includes not just clean data but also rich, machine-readable business context.
- Google’s Agentic Data Cloud aims to provide this foundation through a vertically integrated stack, from custom hardware (TPUs) to data platforms (BigQuery) and AI models (Gemini).
- Gemini Enterprise serves as a simplified, conversational ‘front door’ for users to interact with their data, abstracting the complexity of underlying systems.
- Open standards like Apache Iceberg are crucial for enabling a true multi-cloud and cross-cloud data strategy, allowing data to be queried where it resides without costly movement.
- Google is introducing tools like the Data Agent Kit and integrating services like the Deep Research Agent with the Knowledge Catalog to empower swarms of agents to perform complex, multi-step tasks.
- As AI scales to ‘agent scale,’ managing cost and performance is critical. Google is optimizing every layer of its stack to improve efficiency, citing significant improvements in BigQuery and Spark.
- The future of data and AI development is ‘intent-driven engineering,’ where practitioners focus on high-level objectives and outcomes, while AI agents handle the low-level implementation tasks.