Google DeepMind has launched two new autonomous research agents — Deep Research and Deep Research Max — that can run up to 160 searches and compile expert-grade reports without human intervention. The agents, available through the Gemini API, represent Google’s most ambitious bet yet on fully autonomous AI productivity tools.
Deep Research handles standard multi-step research: it plans a search strategy, executes queries, evaluates results, and synthesizes findings into a structured report. Deep Research Max scales the same capability to deeper, more complex investigations, browsing hundreds of sources and generating comprehensive analyses comparable to professional research reports.
The agents use advanced reinforcement learning to navigate complex information landscapes, adjusting their search strategy as they gather results. Users can set them running overnight and wake up to a completed analysis, a workflow Google is marketing as “research while you sleep.”
The launch places Google in direct competition with similar agentic offerings from OpenAI’s ChatGPT Work and Anthropic’s Claude Cowork. All three companies are racing to define the standard for autonomous AI assistants that go beyond chat to execute sustained, multi-step knowledge work on behalf of users.
