Meta, Stripe, and Sierra Publish Open Standard for AI Agent Interactions
The proposed 'personal agent protocol' aims to establish security and authentication rules as automated software assistants interact with digital businesses.

A coalition of technology and retail companies including Meta Platforms, Stripe, Walmart, and enterprise artificial intelligence startup Sierra has published an open specification designed to standardize how automated software agents interact with commercial websites and enterprise platforms. The initiative, reported by CNBC Business (https://www.cnbc.com/2026/10/06/meta-joins-companies-to-tame-chaos-of-doing-business-with-ai-bots.html), follows the launch of Meta's consumer software agent, Muse, which reached the top of Apple's App Store ahead of ChatGPT after debuting in early September.
Titled the "personal agent protocol," the open framework is intended to resolve ambiguity regarding automated bot activity on business networks. Sierra co-founder and OpenAI chairman Bret Taylor, who is leading the initiative, told CNBC Business that businesses face operational risks when they cannot differentiate between unauthorized web scrapers and legitimate automated assistants acting on behalf of customers. Taylor noted that managing agent access remains chaotic in the absence of a shared industry standard.
Comparing the protocol to single sign-on mechanisms that allow users to log into external services with Facebook or Google credentials—systems Taylor worked on while serving as chief technology officer at Facebook—the standard is intended to establish identity verification and operational transparency. Under the proposed model, businesses can verify incoming agents and observe their actions on digital storefronts.
The publication comes amid friction between online platforms and AI developers over automated access. E-commerce giant Amazon has blocked Meta's software agents from its website over data scraping concerns. Amazon also previously blocked agents from AI search engine startup Perplexity and filed suit against the company in November, alleging Perplexity concealed bot identities to scrape product pages without consent. Perplexity characterized that lawsuit as a bully tactic.
Security and data handling are central components of the specification. David Singleton, head of Meta Superintelligence Labs and former chief technology officer at Stripe, highlighted that consumer agents require access to sensitive information, such as payment credentials and personal details, to complete transactions. Singleton likened the initiative to open email protocols, arguing that industry rails provide essential visibility and access control for both personal and business-facing agents.
Meta reported that Muse has acquired millions of users in the United States and continues to expand rapidly. According to Singleton, users employ the agent to automate everyday tasks, including booking appointments, buying gifts, and registering children for classes, driving the demand for standardized business interfaces.
Neither OpenAI nor Anthropic has formally adopted the specification at launch. Taylor told CNBC Business that he expects major AI developers to participate over time, emphasizing that the long-term viability of the framework relies on its widespread adoption as an open standard.
Sources
Written by
The Company Wire
Inside the companies building what’s next. Reporting on startups, technology, funding and the people shaping them.


