Meta and Sierra have introduced an open protocol to standardise how personal AI agents authenticate with businesses and access services.
Meta and Sierra are developing Personal Agent Protocol, an open standard designed to govern how personal AI agents interact with businesses on behalf of consumers. The initiative includes industry partners Genesys, Instinct, Rocket, Shopify, Stripe, and Walmart, with the aim of establishing a consistent approach to authentication, permissions, and business oversight when AI agents perform tasks for users.
How the protocol works
The protocol is based on the principle that consumers decide what access to give their personal agents, while companies set parameters for what those agents can do. A session starts on the company’s website, where the agent discovers what the company offers and how to reach it. The agent can begin as a guest, which may be enough to check product availability or ask about a returns policy. For tasks requiring account access, the customer can sign in on the company’s page or use credentials already set up with the agent, and decide whether the agent has read-only or write access. Sessions are built on OAuth and carry across channels, so a question asked before sign-in and an order change made afterwards form part of the same visit.
Companies choose the route through which agents complete tasks: their regular web pages, APIs built on standards such as MCP and OpenAPI, or their own agent for tasks that need conversation, such as a warranty claim. Kevin Miller, Head of Payments at Stripe, stated that a customer relationship does not start or end at checkout, and that customers sending an agent expect the same service they would receive themselves. According to Miller, the protocol gives businesses a standard way to recognise and interact with their customers’ agents. Mani Fazeli, VP Product at Shopify, noted that the protocol can help agents and merchants work together across product questions, purchases, returns, and exchanges.
Roadmap and payments extensions
The partners plan to publish the v0.1 specification, host design workshops with interested parties, and release a reference implementation for developers. Potential additions include more detailed permissions allowing customers and companies to set limits on specific actions, push notifications that let a company alert an agent when a flight is delayed or an order ships, and payments extensions that could allow an agent to complete a purchase without sharing credit card information. The latter would extend the protocol directly into checkout flows.
Tony Bates, Chairman and CEO of Genesys, described personal AI as a new front door to the enterprise and stated that brands need a trusted way to know who an agent represents, what it is authorised to do, its intent, and how to work with it securely. Shawn Malhotra, CTO of Rocket Companies, stated that Rocket and Redfin are building agentic technology with Sierra that allows a Muse agent to move across the platform in real time, from finding a home to securing financing. The emphasis on both consumer permissions and brand-side control reflects the partners’ view that adoption at scale depends on the connection working for consumers, brands, and agent builders alike.
