Google has introduced a redesigned version of its Gemini Deep Research agent, powered by the new Gemini 3 Pro model, marking a significant step toward developer-ready agentic AI. Using the Interactions API, developers can now integrate Google’s search systems directly into their own applications, enabling large-scale synthesis in areas such as due diligence and pharmaceutical safety. Google plans to integrate the agent into Search, Finance, the Gemini app and NotebookLM, positioning AI agents (not users) as the future interface for searching complex information. The launch emphasizes the improved factual accuracy of Gemini 3 Pro, supported by new benchmark tests, including Google’s open source DeepSearchQA. Early comparisons show Google outperforming competitors in several tests, although competition intensified on the same day with OpenAI’s release of GPT-5.2.
Along with the product rollout, Google is facing a copyright challenge from Disney, which issued a cease and desist letter alleging that Google’s AI services were generating and distributing unauthorized images of major Disney characters. Disney claims that the releases imply endorsement due to the Gemini brand appearing on the generated content. Google said it would engage with Disney while noting the copyright controls already built into its platforms.
Google has also expanded its virtual try-on technology, which initially launched in July. Users can now view clothes in search, purchases and images with a dedicated “try on” option. The company’s Doppl app has been updated with a shoppable discovery feed featuring AI-generated outfit videos and direct links to merchants, aiming to deliver a familiar, social media-inspired shopping experience powered by generative AI.
