Google and Meta have begun internal trials of personal AI agents codenamed Remy and Hatch, according to reporting published May 6, 2026. Both efforts are positioned to act autonomously on users’ behalf for everyday tasks and integrate more deeply into core product surfaces, a shift framed as a response to the commercial head start achieved by competitors.
Remy is operating inside an employee version of Google’s Gemini app and is designed to connect with other Google services. Internally, it is described as a round‑the‑clock personal agent for work, school and daily life that can answer questions, carry out tasks on its own, proactively monitor relevant activity, and learn user preferences over time. Google declined to comment on the tests.
On May 4, 2026, Google announced it would shut down Project Mariner, its experimental browser agent, and fold the technology into Gemini Agent. Reports say staff who had worked on Mariner were reassigned to focus on an OpenClaw response, and agents are expected to be a major focus at Google’s I/O event scheduled this month. Meta’s Hatch is being developed alongside an agentic shopping tool for Instagram and is slated to enter internal testing by the end of June. The Hatch prototypes have been trained in sandboxed web environments that simulate sites such as DoorDash, Etsy and Reddit to prepare the agent for real‑world interactions with commerce and social content.
So far Meta has used Anthropic’s Claude models to train Hatch but plans to switch the agent to its own Muse Spark model at launch. The Instagram shopping feature is planned for integration before the fourth quarter and would let users tap a product in a Reel to learn more and complete a purchase without leaving the app, a direct play to compete with TikTok Shop.
Industry observers say Anthropic and OpenAI retain a commercial lead. Anthropic currently offers products such as Claude Code and Claude Cowork; OpenAI has added OpenClaw creator Peter Steinberg to its team and is building agent‑capable products alongside Codex and a reported “super app.” Microsoft has also integrated Anthropic technology into Copilot Cowork. The broader takeaway is that the standalone browser‑agent wave has softened, and major platforms are now expected to embed autonomous assistants into email, calendars, office suites and commerce surfaces. Companies are prioritizing agent development as they seek returns on recent AI investments and scale up infrastructure spending this year.
Sources
Replies (0)
No replies in this topic yet.
