Google adds natural language query capabilities to AlloyDB

Google is enhancing AlloyDB, its fully managed database-as-a-service (DBaaS), to help developers build applications underpinned by generative AI.

Announced at Google’s annual Cloud Next conference, the updates could give the PostreSQL-compatible AlloyDB an edge over PostgreSQL itself or other compatible offerings such as Amazon Aurora.

[ Related: Google Cloud Next ’25: News and insights ]

Among the additions, a new AlloyDB AI query engine enables developers to use natural language expressions inside SQL queries.

When Google launched AlloyDB in May 2022 the open-source PostgreSQL was rising in popularity due to its transactional and analytical capabilities, extended support for spatial data, broader SQL support, enhanced security and governance, and expanded support for programming languages. Google saw an opportunity to offer a cloud-based alternative as a service — but it’s an opportunity that also attracted the attention of rival Amazon Aurora and Microsoft Azure’s Database for PostgreSQL. Now the challenge for Google is to make its offering stand out.

Support for natural language in SQL queries

The arrival of the AlloyDB AI query engine means that developers can now embed free text questions inside SQL queries, even if those depend on less-structured data such as images and descriptions, said Bradley Shimmin, lead of data and analytics practice at The Futurum Group.

This, said ISG Software Research’s executive director David Menninger, will ease the burden on developers as they need to be very precise when writing SQL statements.

By way of example, Menninger said, instead of writing “SELECT customer_name FROM customer_table WHERE city in (‘Boston’, ‘Cambridge’)”, a developer using AI Query could give a narrative description of what they were looking for, such as “list all the customers near the Charles River.”

With its new query engine, said Futurum Group’s Shimmin, Google is following the trend of database providers converging database operations with semantic and relational query methodologies to expand capabilities of traditional SQL use cases.

Alongside the query engine, Google is adding the next-generation of AlloyDB natural language capability that is expected to allow developers to query structured data inside AlloyDB, thereby helping them build applications that understands an end user’s natural language input better.

ISG’s Menninger sees the natural language capability as a productivity tool for developers.

“It’s often easier to write a natural language query than to write out a SQL statement. It may not be the final SQL statement, but what’s generated can be edited so it moves the development process along more quickly,” Menninger said.

For the enterprises, the analyst sees the natural language ability making data more accessible for end users.

“You don’t necessarily need an analytics tool. You can simply ask the database some questions and get responses. And developers can embed these capabilities in the applications they create benefitting end users,” Menninger explained.

Google Agentspace can now search structured data in AlloyDB

As part of the updates to AlloyDB, Google said that enterprises that subscribe to its Agentspace service will now be able to search structured data inside AlloyDB.

The Agentspace service, launched in December, is intended to help enterprises build agents, which in turn can be used to search data stored across various sources.

These agents can also be programmed to take actions based on the data held at different sources within an enterprise, the company said.

dbInsights’ chief analyst Tony Baer said the extension of Agentspace to AlloyDB is a logical move as Google expects that enterprises will use agents to work or interact with their data in the future.

More support for migrations and other updates

Other database updates announced at Cloud Next this year includes updated support for migrations, added support for running Oracle’s Base Database service, and Model Context Protocol (MCP) support for Gen AI Toolbox for Databases.

The update to migrations comes in the form of the Google Database Migration Service (DMS) now supporting SQL Server to PostgreSQL migrations for Cloud SQL and AlloyDB.

“These new capabilities support the migration of both self-managed and cloud-managed SQL Server offerings, and a range of SQL Server editions and versions,” Google said in a statement.

In April last year, Google added Gemini support to the DMS to make migrations faster by supporting conversion of database-resident code.

The Toolbox is an open-source server designed to streamline the creation, deployment, and management of AI agents capable of querying databases.

With industry support for MCP rising, it was inevitable that Google would add support for the protocol to its GenAI Toolbox for Databases, analysts said.

Anthropic introduced MCP last year to make it easier to bring data to LLMs. Since then, it has become a standard means of linking up models, tools, and data resources in support of agentic processes, Shimmin said.

For Menninger, MCP is the emerging standard that enterprises are starting to use to provide context to agents in order to enhance their performance.