Your Customer's Agent Is at the Door: The Personal Agent Protocol and the 'Agent Front Door' for Businesses

So far, the enterprise agent debate has mostly centered on 'how do we manage our own agents?' The new question comes from the other direction: what happens when customers' own AI agents come to your company to place an order, ask about a return or change an account?

What is the Personal Agent Protocol?

• On October 6, Sierra and Meta announced an open standard defining how users' personal AI agents authenticate with businesses and what scope they act under.
• Partners include Walmart, Shopify, Stripe, Rocket, Genesys and Instinct.
• Sessions are built on OAuth: an agent can start as a guest to check stock or a returns policy; once the customer signs in, they decide whether the agent gets read-only or write access.
• The business chooses whether agents reach it through its website, through APIs such as MCP and OpenAPI, or through its own agent.
• Payments are not in the first version; they are planned as a future extension.

Still early

The v0.1 specification text has not been published yet; it is expected later in October and there is no reference implementation so far. OpenAI, Anthropic and Amazon have not joined for now, while Stripe and Shopify are also part of Visa's rival Trusted Agent Protocol. The standards battle is only beginning.

Four preparations for companies

• Know your agent traffic: measure how much of the automated traffic to your site and APIs comes from AI agents.
• Define scope up front: decide what a guest agent can see and what a signed-in customer's agent can change.
• Open a separate door for agents: having agents arrive through a structured API rather than by scraping screens serves both you and the customer.
• Log every action: which agent did what on whose behalf should be demonstrable if a dispute arises.

The remot angle

What answers incoming customer agents will often be the company's own AI assistant. remot manages the model calls behind that assistant from a single point: it limits which data can be reached according to the incoming agent's permission scope, masks personal data and logs every interaction. That way opening the door to customer agents doesn't mean losing control.