Verified, then logged.
Sun 22 Nov 2026 · 00:34 KST / 15:34 UTC한국어 | ENSearch /

01 / AI · Analysis

Google previews two AI text functions for AlloyDB

Google puts classification and extraction into SQL, while processing still includes sending inputs to models.

Image from Google
Image: Google

Google announced a preview of two AI functions in AlloyDB, its database service. That count comes from the announcement, which provides no throughput measurements.

The functions bring text classification and extraction into SQL, the language used to query databases. Google said they were designed to reduce the need to write separate model instructions.

Google also described grouping inputs before sending them to models. Its account of processing inside the database therefore includes a model transmission step.

Text processing through SQL

According to Google, ai.classify categorises text using categories defined by the user.

Google described ai.extract as extracting values for specified fields. It returns JSON, a structured data format.

The announcement’s examples require a preview setting to be enabled. The classification example presents its results as expected output.

Performance claims behind batching

Google said the database groups AI inputs used in SQL WHERE clauses, which set conditions for a query. It then sends those batches to models on a model platform.

Google said batching improves speed and reduces costs. The announcement provides no comparative figures showing the size of those improvements.

Google outlined plans to extend processing beyond WHERE clauses and add provenance tracking, which records where output comes from. Those remain planned additions to the preview’s stated scope.