During Google Cloud Next, Andi Gutmans stated that Google Cloud possesses a structural advantage in the enterprise AI agent market. He argued that no competitor currently integrates cloud computing infrastructure, frontier AI models, and a data platform under one roof.
Gutmans explained that while AWS and Azure offer infrastructure, they lack proprietary frontier models. Data providers have data platforms but rely on external infrastructure and models. AI model providers, conversely, specialize solely in models. He stressed that Google Cloud's integrated stack is crucial for realizing the full value of AI.
As enterprises transition from human-query-responsive AI tools to autonomous agents, the significance of these component gaps becomes more pronounced. This shift places increased pressure on underlying data platforms and rewards providers controlling more of the technology stack for cost-effective agent operation at scale.
Gutmans noted that Google spent the last 18 months redesigning its data platform for agent scale. He highlighted that approximately 90% of enterprise data remains unstructured and historically underutilized. The newly announced Knowledge Catalog aims to make this data accessible to agents without requiring extensive manual preparation by data engineers.
The pivotal moment enabling this transformation was a model-level advancement rather than a product decision. The arrival of Gemini 2.5 marked a turning point in reasoning capability, compelling Google to re-engineer every agent within its data portfolio.
He elaborated that Google completely re-engineered all its agents—including conversational analytics, data science, and data engineering agents—in the past year, making them less prescriptive with models. The Knowledge Catalog and MCPs significantly aid this by enhancing reasoning capabilities. Gutmans exemplified this by noting a dramatic improvement in customer usability for conversational analytics compared to the previous year.