Meta’s personal AI assistant, Muse, has vaulted to the top of the App Store and contributed to a more than 20% rise in the company’s stock price since its launch this month. The product, unveiled at Meta’s Connect conference, is being promoted as a flagship result of the firm’s multibillion‑dollar AI investment.
Intensive development and internal testing
The effort behind Muse spanned months of “blood, sweat, and tears,” according to employees familiar with the project. Nat Friedman, former GitHub chief executive who now heads Meta’s Superintelligence unit’s Products and Applied Research team, led the initiative. Friedman, who joined Meta after a deal involving a stake in his venture fund NFDG, oversaw a team that included recent acqui‑hires from the AI‑agent startup Dreamer.
Early internal trials revealed significant shortcomings. Two weeks before launch, the chatbot still struggled with fluent English and was described by a tester as feeling “like just another chatbot, but with a worse underlying AI model and extra steps.” Tens of thousands of Meta employees subsequently tested Muse for months, providing use‑case data that helped the team improve tasks such as booking meetings and canceling plane tickets – a process known in Silicon Valley as “hill‑climbing.”
Friedman’s hands‑on style pushed the team to discard slow internal tools in favor of faster external solutions like Vercel. He also responded quickly to the popularity of the open‑source AI agent OpenClaw, purchasing hundreds of Mac Minis for the superintelligence team and issuing an internal memo titled “The Claw Is The Law.”
Market impact, hiring spree, and future outlook
At Connect, AI chief Alexandr Wang, 29, called Muse a “banger” and highlighted the research team’s work teaching agents to select appropriate tools for tasks. The app’s rapid climb on the App Store was aided by Meta’s massive built‑in audience across its platforms, reaching users who had never previously tried an AI agent.
Meta’s hiring surge reinforced the development push. The company brought in Dreamer co‑founders David Singleton and Hugo Barra, along with their team, to strengthen agentic AI capabilities. Singleton, who joined in March, noted his work on the core technology that secures sensitive user information.
Analysts have taken note. Mizuho Americas analyst Lloyd Walmsley wrote that Muse “shows the Meta bull case increasingly playing out.” Bank of America analysts, however, warned that privacy and trust remain “key roadblocks” as the app requires broad access to user accounts and could make purchases without explicit permission.
CEO Mark Zuckerberg has said the long‑term goal is to enable shopping and savings through Muse, with Meta taking a small cut of each transaction. He also disclosed that he and Friedman consulted Signal creator Moxie Marlinspike to co‑design Muse’s security architecture, citing safety and product‑liability concerns as reasons for delaying the launch.
Looking ahead, Meta hinted at a new AI model called “Watermelon,” which the company claims has already caught up with OpenAI. Wang posted that “the kitchen is just getting warmed up,” suggesting additional large‑scale AI initiatives are on the horizon.
For employees still coping with a year of layoffs, mass surveillance, and low morale, Muse’s success offers a morale boost and a tangible sign that Meta remains competitive in the AI race after being perceived as a laggard a year ago.
Mitchell Landsberg is a Senior Technology Correspondent at News Raise. He covers consumer electronics, artificial intelligence, software developments, and digital privacy trends.
