Meta's personal AI agent Muse has topped the U.S. App Store free app chart about 10 days after launch, surpassing ChatGPT. Instead of staking security on a model that is "obedient enough," it gives each user a separate cloud computer, uses Sentinel to approve network and third-party service access, and isolates payment details with one-time payment cards. As agents begin handling tasks on users' behalf, permissions and boundaries are becoming central to whether a product can be trusted—an...

Muse topped the U.S. App Store free app chart about 10 days after launch, beating ChatGPT. Almost simultaneously, Manus, another representative product on the personal AI agent track, was reported to be raising a new round: $500 million at a valuation of about $4 billion.

AI agents are giving frontier AI giants a path to "overtake on the curve." Foundation models still matter, but on the agent track, product design, tool calling, task execution, permission management and user entry points all enter the battlefield. Latecomers do not necessarily have to beat OpenAI and Anthropic on model leaderboards first; they can also win at the application layer first.

When Muse launched, its features alone did not seem especially unique. It can browse the web, call tools, process files, and book hotels, send emails and shop for users. ChatGPT, Claude and the earlier Manus have all demonstrated such capabilities. What makes