Muse arrived as Meta’s consumer-facing AI agent and quickly climbed to No. 1 in the US App Store, with an estimated 600,000 daily active users in the country. But as the launch gathered pace, users began comparing its inner workings with OpenClaw, the open-source agent platform that had already won over much of the AI world.¹

The allegations were specific, not merely stylistic. Users highlighted shared workspace file names including SOUL.md, memory and tools; similar product design; and personality instructions containing lines such as “Be genuinely helpful, not performatively helpful.” One critic summarized the suspicion bluntly: Muse was “just a wrapper app built on top of” OpenClaw.²

Meta’s response did not deny the inspiration. Nat Friedman, product head at Meta Superintelligence Labs, said Muse was “built from scratch” but “definitely heavily inspired as a product by OpenClaw.” He said that after trying OpenClaw in January, he bought hundreds of Mac Minis for the MSL team, whose members became enthusiastic users of the tool.¹

That admission sharpened, rather than settled, the dispute. When challenged over Muse’s identical file names and nearly identical SOUL.md content, Friedman said OpenClaw creator Peter Steinberger “got those things exactly right.” Meta’s case is that it borrowed a successful product blueprint, then rebuilt it for safety, simplicity and global scale. Its stated aim was to create something like OpenClaw that could be “safe and secure and easy to use and scale to billions of people.”¹

For skeptics, however, the repeated matches make “from scratch” sound less like a clean break than a carefully drawn corporate boundary. Meta has offered no further comment beyond Friedman’s explanation.²