Personal AI agents that book appointments, shop and fill forms for people are useful—and messy for the companies whose websites they land on. On 6 October 2026 CNBC reported that Meta, Walmart, Stripe and Bret Taylor’s enterprise AI start-up Sierra published an open “personal agent protocol” meant to set how such agents interact with businesses. [1]
What problem are they trying to solve?
Taylor, who co-founded Sierra, chairs OpenAI and is leading the initiative, told CNBC that companies still struggle to know when a visitor is a personal agent acting for a person versus a random bot. “It is kind of chaos until such a standard exists,” he said. [1]
The protocol is pitched as an open standard so businesses can authenticate personal agents and gain visibility into what those agents do on their sites—Taylor compared it to logging in with Google or Facebook credentials, work he did as Facebook’s tech chief. [1]
Why Muse made this urgent
Meta’s Muse personal agent launched in early September 2026, soared to the top of Apple’s App Store ahead of ChatGPT, and became the most discussed personal-agent product in CNBC’s framing. Amazon has blocked Meta’s agents over scraping concerns; Amazon also blocked Perplexity’s agent and sued Perplexity in November (year not restated in the piece), alleging concealment of AI agents—claims Perplexity called a “bully tactic.” [1]
David Singleton, vice president of engineering and consumer products at Meta Superintelligence Labs, said Muse already has millions of US users and is “growing rapidly.” He listed everyday uses—signing kids up for classes, buying gifts, scheduling dentist appointments—and argued customers must share payment and personal data for agents to work well, so rails for visibility and control matter. He compared the goal to email: a shared standard everyone can use. [1]
Who is in—and who is not yet
CNBC names Meta, Walmart, Stripe, Sierra and “a handful of other companies” as publishers of the protocol. OpenAI and Anthropic are not on board yet; Taylor said he expects them to join an open standard and would be “really disappointed” if competitors do not use it. [1]
What we don’t know
- Technical spec depth. CNBC describes intent (authentication, visibility, security) but does not publish the protocol’s full technical document.
- Adoption. Whether major retailers beyond Walmart, or rival agent makers, will implement it remains aspirational in the reporting.
- Muse scale. “Millions” of US users is Singleton’s company-stated figure, not an audited metric.
- Whether blocking ends. Amazon’s blocks and the Perplexity dispute show merchants may still prefer hard gates until standards prove themselves.
The Bottom Line
Meta, Walmart, Stripe and Sierra are proposing open rails so shops can tell a person’s AI agent from a scraper bots—and see what the agent does—after Muse’s viral launch collided with merchant blocks. It is an industry governance bet, not a finished universal login; OpenAI and Anthropic are not signed up yet, and Muse’s “millions” of users are Meta’s claim.
Sources
- CNBC, Kate Rooney, 6 October 2026 — “Meta joins with group of companies to tame ‘chaos’ of doing business with AI bots”: https://www.cnbc.com/2026/10/06/meta-joins-companies-to-tame-chaos-of-doing-business-with-ai-bots.html
