Google outlines agentic, inexact-query database future

The Register's coverage of Google Cloud Summit in London records on-the-record comments from Sailesh Krishnamurthy, VP engineering, and Yasmeen Ahmad, product executive for Agentic Data Cloud. Ahmad is quoted saying, "we're putting agents at the center... with the goal that humans are not going to be using data platforms in the next three to five years. It's going to be humans orchestrating agents, and agents actually doing the work." Krishnamurthy is quoted describing a shift where, when retrieving data, "it's not so much about getting the exact results, but getting the best results." The Register reports Google is investing in a knowledge catalog (formerly Dataplex) and treating enterprise search as LLM context that aggregates structured and unstructured organization data.

The quotes and reporting emphasize vector indexing, text indexing, and graph technology as core elements of what Krishnamurthy calls "AI native infrastructure." The Register records Krishnamurthy noting combinations of structured and unstructured data will be operated in terms of inexact results and data quality. The coverage also preserves Krishnamurthy's caveat that exact SQL queries "are not going away" and that some natural-language fuzzy queries may still generate exact SQL when needed.

Public reporting frames this as part of a broader move across cloud and database vendors toward supporting retrieval-augmented and agentic workflows. Companies integrating vector search, knowledge graphs, and catalog services typically surface more contextual signals for LLMs, while still retaining deterministic query paths for compliance and verification. For practitioners, this pattern raises implementation tradeoffs around indexing cost, latency, and testability when combining probabilistic retrieval with exact query execution.

Observers should track product announcements and documentation for how Google exposes verification and explainability for AI-generated queries, the latency and cost profiles of hybrid indexing stacks, and how the knowledge catalog evolves to provide provable lineage and schema mappings for LLM context. The Register story does not quote a Google roadmap beyond the product-exec remarks, and no additional company statements on rationale were provided in the coverage.