QueryData watch: signs

Google did not name QueryData, but its latest Deep Research rollout points in the same direction: cloud agents that turn messy questions into structured research jobs. (blog.google) On April 21, 2026, Google introduced Deep Research and Deep Research Max, two agents built with Gemini 3.1 Pro for longer-running research tasks. Google said the new versions can search the web, connect to custom data through Model Context Protocol, and return cited analyses. (blog.google) Google’s developer documentation says the Deep Research agent “autonomously plans, executes, and synthesizes” multi-step research tasks and runs through the Interactions application programming interface, not the standard `generate_content` path. The docs also say the agent works asynchronously in the background and can combine public web results with a user’s own files. (ai.google.dev) That setup is the basic shape of a query agent: software that takes an open-ended request, breaks it into searches, reads sources, finds gaps, and searches again. Google described that loop in December 2025 when it first opened Gemini Deep Research to developers through the Interactions API. (blog.google) Google’s newer model release fills in the other part of the picture. When Google announced Gemini 3.1 Pro on February 19, 2026, it said the model was designed for advanced reasoning tasks such as synthesizing data and explaining complex topics, and that it was rolling out across the Gemini API, Vertex AI, the Gemini app, and NotebookLM. (blog.google) Google has also been moving the same capabilities into products with much larger distribution. In November 2025, the company said Gemini 3 would ship in Search on day one and described the model as better at figuring out the “context and intent” behind a request with less prompting. (blog.google) The Deep Research launch adds more concrete plumbing for that intent-to-query workflow. Google said one API call can now trigger research that blends the open web with proprietary data streams, and it split the product into a faster agent for interactive surfaces and a slower “Max” version for background jobs such as overnight analyst reports. (blog.google) Google’s December 2025 post used nearly the same framing, saying the agent formulates queries, reads results, identifies knowledge gaps, and searches again. That is not a direct QueryData announcement, but it is a clear sign that major cloud platforms are productizing the same underlying behavior. (blog.google)