Google Cloud Next '26 Notify me• Apr 23, 2026• 20:19ConferenceFireside Chat

From systems of intelligence to systems of action: Yasmeen Ahmad on the agentic data cloud

From Google Cloud Next '26 · 2026

Jason Davenport(Moderator, Google Cloud)•Stephanie Wong(Host, Google Cloud)•Yasmeen Ahmad(Managing Director, Data Cloud, Google Cloud)

Get the full transcript next time Google Cloud Next '26 releases an episode

Summary, key quotes, top claims, and the searchable transcript - emailed automatically. No card needed.

Executive Summary

Continue your research

Keep pulling the thread on Yasmeen Ahmad.

From Intelligence to Action The Primacy of Business Context

9 quotes

Concerns Raised

The historical failure to operationalize the majority of data insights.
The potential for exponential cost increases as agent-driven workloads scale.
Data silos across multiple clouds and platforms hindering a unified view.
The inadequacy of traditional data quality metrics for achieving high-accuracy AI agents.

Opportunities Identified

Automating complex business processes with 'swarms of agents' to achieve dramatic time savings (e.g., 45 minutes to 1 minute).
Unifying disparate data sources across clouds using open standards like Apache Iceberg.
Leveraging AI to automatically infer and codify business context from unstructured data at scale.
Achieving significant price-performance gains through a vertically integrated AI stack.

Key Themes

Research Findings12

According to Google Cloud's experience, focusing only on traditional data quality metrics like cleanliness and lineage results in AI agents achieving only 50% accuracy.

Google Cloud has integrated its deep research agent with the Knowledge Catalog, enabling it to reason over a combination of enterprise data, web data, and documents.

Google Cloud launched the Data Agent Kit at Google Cloud Next '26, which provides plugins, extensions, tools, and skills for AI agents.

Google Cloud's cross-cloud solution can connect to data in Databricks via Unity Catalog, Snowflake via Polaris, and AWS S3 via Glue.

Over the last year, Google Cloud has improved BigQuery's query processing speed by 35% while simultaneously reducing its cost by 40%.

Google Cloud's managed service for Apache Spark, using the new Lightning engine, is 5 times faster than standard Apache Spark.

Google Cloud's managed service for Apache Spark with the Lightning engine offers 2 times better price-performance than the leading proprietary market alternative.

Running AI inference over data in BigQuery results in a 230x reduction in token usage due to Google Cloud's integrated stack optimizations.

Google Cloud customers are reporting that using swarms of AI agents has reduced the time for certain human processes from 45 minutes down to one minute.

Historically, only an estimated 10% to 20% of data insights generated by corporate data platforms were successfully put into production to drive business actions.

Google Cloud is developing an "agentic data cloud" designed to translate generative AI insights into direct business actions.

The remaining 50% of AI agent accuracy, beyond data quality, is derived from providing the agents with business context.

Topics

AI Agents Agentic Data Cloud Systems of Action Google Cloud BigQuery Apache Spark Cross-Cloud Multi-Cloud Apache Iceberg Data Federation Knowledge Catalogue Intent-Driven Engineering Generative AI Data Strategy Price-Performance

Processed Apr 28, 2026Daily intelligence brief → yt-dlp + mlx-whisper + Gemini

From systems of intelligence to systems of action: Yasmeen Ahmad on the agentic data cloud

Continue your research

Concerns Raised

Opportunities Identified

Key Themes

From Intelligence to Action

The Primacy of Business Context

The Rise of Agent Swarms and Intent-Driven Engineering

Open, Cross-Cloud Data Federation

Full-Stack Optimization for Agent-Scale Workloads

Research Findings12

Topics