Google Cloud expands Duet AI preview across developers and data teams

Managing a complex cloud infrastructure usually means swimming through endless dashboards and writing cryptic query syntax just to figure out why a server crashed. Google wants to replace all of that with a simple chat prompt.

At Google Cloud Next, Google announced a massive expansion of Duet AI across its entire cloud ecosystem, with plans for general availability later this year. What started at Google I/O in May as a coding assistant is now reaching into operations, data engineering, database migrations, and cybersecurity.

Across the entire developer stack

Duet AI is moving beyond standard code completion. Beyond integrating into third-party IDEs like VSCode and JetBrains, Google showed off how Duet AI can modernize legacy code. In one demo, a developer asked the AI to rewrite a C++ catalog service into Go and rewire it for Cloud SQL in a single prompt. Engineering platform Turing reported seeing productivity gains of around 33 percent during early trials.

Google is also using Duet AI to tackle painful database migrations. Through its Database Migration Service, the assistant can automate the conversion of complex stored procedures and PL/SQL code, clearing the path to migrate legacy Oracle databases to Cloud SQL or AlloyDB later this year.

For data teams, BigQuery users can now use conversational prompts to generate SQL and Python, run vector embeddings, or call Vertex AI foundation models with a single query. Business intelligence tool Looker is also gaining natural language capabilities to turn raw data into slide decks and summaries.

Turning rookie analysts into security pros

Google is also putting its 5.4 billion dollar acquisition of Mandiant to work. Security teams are constantly overwhelmed by threat alerts, so Google embedded Duet AI into Chronicle Security Operations, Mandiant Threat Intelligence, and Security Command Center.

Instead of mastering complex query languages, junior security analysts can ask plain-English questions to evaluate potential threats, map attack paths, and get response recommendations. Addressing concerns about AI hallucinations in high-stakes security work, Vertex AI product lead Nenshad Bardoliwalla noted that Google is grounding the models with authoritative security data sources to control accuracy.

Why it matters

For engineers and data practitioners, Duet AI looks genuinely useful because it cuts down context switching across the console. Translating natural language into PromQL, SQL, or log debugging saves real hours every week. If your company is already running on Google Cloud or planning an expensive database migration, testing these preview features is an easy call.

Across the industry, this is Google's most direct counterpunch yet to Microsoft's Copilot strategy. Instead of treating AI as just a code-completion add-on, Google is weaving it directly into proprietary crown jewels like BigQuery and Mandiant intelligence. It makes the entire Google Cloud ecosystem stickier.

That said, we think the biggest hurdle remains trust in high-risk zones. While a buggy code suggestion is caught by a compiler, a hallucinated security assessment or a flawed database migration script can cause real damage. Grounding helps, but enterprises will likely treat Duet AI as an adviser rather than an autopilot for quite some time.