From Search Box to Autonomous Google Search Agents
Google is redefining what a search engine does by introducing Google Search agents that actively work on users’ behalf. Instead of simply returning links, these agents can continuously scan web sources and Google’s real-time datasets for topics, listings, or products that match user-defined requirements. Early access will come through Google AI Pro and Ultra subscriptions, positioning agents as a premium layer on top of traditional search. Under the hood, the company is upgrading Google AI Mode with Gemini 3.5 Flash as the default model, after AI Mode surpassed one billion monthly users and saw query volumes more than double every quarter since launch. The revamped experience also starts with a redesigned, multi-modal search box that accepts text, images, files, videos, and even Chrome tabs, then carries context forward into AI conversations. Together, these changes transform search from a one-off lookup into an ongoing, agent-powered service.
AI Mode, Generative Interfaces, and Search Automation
Google AI Mode is evolving into a workspace for complex, multi-step AI search tasks. Users can move from AI Overviews into conversational mode, refine their intent with follow-up questions, and maintain context while still seeing supporting links and articles. Beyond text answers, Search is gaining generative interfaces that assemble custom layouts—tables to compare options, graphs to visualise trends, dashboards and trackers to monitor evolving topics. These generative UIs make agent output more like interactive tools than static results. Google is also extending Personal Intelligence in AI Mode to nearly 200 countries and territories across 98 languages, letting users connect Gmail and Google Photos, with Calendar integrations planned. Combined with emerging agent features, this turns Search into a proactive environment that can remember user constraints, reference personal data where permitted, and present information in task-oriented formats, signalling a clear move toward search automation rather than mere information retrieval.
Universal Cart: When AI Search Tasks End in a Purchase
Search agents are tightly linked to a new commerce layer through Universal Cart, Google’s cross-service shopping feature. As users browse in Search, the Gemini app, YouTube, or Gmail, they can add items into a single cart that is continuously monitored by Gemini models. The system flags price drops, shows price history, tracks stock changes, and surfaces deals across merchants, helping agents optimise when and where to buy. Built on Google Wallet, Universal Cart factors in payment benefits, loyalty programmes, and merchant offers at checkout, and can even warn about compatibility issues, such as mismatched PC components from different retailers. Purchases are executed via the Universal Commerce Protocol, which Google describes as a common language for agent-based commerce. Shoppers can either pay with Google Pay through participating brands or hand off to a merchant site, allowing search-driven agents to move seamlessly from discovery to transaction.
Booking, Payments, and the Rise of Agent-Led Execution
Beyond shopping, Google is extending agentic capabilities into service booking and payments, deepening the shift to search-as-execution. In Search, users will be able to specify detailed requirements for local experiences or services, then receive pricing and availability alongside links to finish the booking. In certain categories like home repair, beauty, and pet care, Google’s systems can even call businesses on the user’s behalf to help complete arrangements. On the payments side, the Agent Payments Protocol (AP2) will allow AI agents to pay within user-defined limits, such as authorised brands, products, or spending caps. AP2 creates a verifiable connection between user, merchant, and payment provider, backed by digital mandates that record the user’s instructions. Together with Universal Cart and UCP-powered checkout, these tools make it possible for Google Search agents to not only recommend actions but to carry them out, with explicit guardrails around autonomy and accountability.
Content Verification and Trust in AI-Driven Search
As AI becomes more embedded in search automation, Google is investing heavily in verification to keep AI-generated results trustworthy. Its SynthID watermarking system has already been used on more than 100 billion images and videos and the equivalent of 60,000 years of audio, allowing tools to later detect whether content was created or edited by AI. Google is now bringing SynthID verification into Search itself, alongside Lens, AI Mode, Circle to Search, and the Gemini app, with Chrome integration coming soon. The company is also expanding support for C2PA Content Credentials, which disclose how media was captured or modified, including whether AI tools were involved. On Google Cloud, an AI Content Detection API will help enterprises identify AI-generated media for use cases such as feed ranking, fraud checks, and fact-checking. These measures aim to ensure that as Google Search agents automate more tasks, users can still verify the integrity and origin of what they see.