Google, Meta Test Personal AI Agents to Close Gap With Rivals
Google and Meta are internally testing personal AI agents designed to handle everyday tasks autonomously, The Decoder reported Tuesday.
The agents — codenamed “Remy” at Google and “Hatch” at Meta — are designed as integrated AI assistants that can operate across multiple services on behalf of users, from managing email and calendars to handling shopping tasks, according to the report. The two companies are working to match agentic AI capabilities developed by Anthropic and OpenAI.
Google has shut down its Mariner browser agent project to concentrate resources on the new approach. Rather than building a standalone browser-based agent, the company is pursuing assistants embedded directly into its existing product ecosystem, according to the report.
The moves come as Anthropic and OpenAI have expanded their agentic AI capabilities. Anthropic’s Claude offers computer use and coding agent features, while OpenAI has expanded its agent offerings across consumer and enterprise products. Both companies have shipped agentic tools that can execute multi-step tasks with minimal human intervention.
The competitive dynamics reflect a broader industry shift toward AI agents — software that can plan, reason, and take actions rather than simply responding to prompts. All four companies are headquartered in the United States, and their product launches are expected to directly shape the AI tools available to American consumers and businesses.
For Google, the decision to shut down Mariner signals a bet that tightly integrated agents woven into Gmail, Google Calendar, and Google Shopping will prove more valuable than a general-purpose browser automation tool. The approach leverages Google’s existing user base across its suite of services.
Meta’s Hatch project suggests the social media company sees personal AI agents as a key battleground beyond its current AI assistant features embedded in WhatsApp, Instagram, and Facebook.
Companies are competing not just on model intelligence but on the ability to reliably perform real-world tasks. The shift from chatbot-style interactions to autonomous task completion has become a central focus for major AI providers seeking to differentiate their products and justify growing infrastructure investments.
