Skip to header Skip to main navigation Skip to main content Skip to footer

Meta and Sierra Draft the Personal Agent Protocol, an Open Sign-In Standard for Agents

Breadcrumb

  • Home
  • Meta and Sierra Draft the Personal Agent Protocol, an Open Sign-In Standard for Agents
Cartoon illustration of a robot agent presenting a passport-style ID card to a shopkeeper at a storefront checkout while a scanner glows.

Your AI agent is about to walk into a store. Meta and Sierra want to make sure it knows how to knock.

On October 6, at Sierra Summit in San Francisco, Sierra and Meta announced the Personal Agent Protocol — PAP, an open standard for how a person's AI agent signs in to a business and gets a task done there. Sierra co-founders Bret Taylor, the former Salesforce co-CEO, and Clay Bavor published the announcement with founding partners Walmart, Stripe, Shopify, Genesys, Rocket, and Instinct.

The problem it solves is real. Today, an agent typically acts like a human visitor: clicking through pages, hoping the form works. "It is kind of chaos until such a standard exists," Taylor told CNBC. PAP gives the agent a proper door: sessions built on OAuth, the web's common sign-in standard. An agent can start as a guest for light tasks like checking stock; for account tasks, the customer signs in and chooses read-only or write access. The business decides what agents are allowed to do, and picks one of three routes: its existing website set up to be read by agents, its APIs over standards like MCP or OpenAPI, or its own conversational agent for tasks like a warranty claim.

The spec itself is still cooking. Sierra plans to publish v0.1 later in October, with a reference implementation and design workshops to follow. On October 9, Sierra published a draft nicknamed Poppy — companies publish one file at /.well-known/poppy.json — and named 35 more design partners, OpenAI among them. That is a notable evolution: OpenAI and Anthropic were conspicuously absent from the original announcement.

Payments are explicitly a future extension, not part of v0.1. Agent checkout is being addressed separately — by OpenAI's Agentic Commerce Protocol and Visa's Trusted Agent Protocol, among others. PAP answers "who is this agent and what may it do here?" and leaves "how does the money move?" to someone else.

The field is crowded — PAP, UCP, ACP, and TAP all cover different slices of the transaction, and the same companies keep appearing on several of them. But PAP's positioning is clever: it is less a rival than a layer beneath the others. If it gains traction, it becomes the plumbing through which personal agents browse, book, and interact with services on a user's behalf.

Watch the spec drop later this month. Standards wars are won by whoever ships the boring, working file first.

Rocky Mountain Madman LLC
AI Implementation
Meta
AI Product News
Enterprise AI
AI Policy

Ai Industry News

Goodfire's Inside-Out Monitors: Reading an Agent's Mind Beats Reading Its Transcript

Cartoon illustration of an airport-security-style scanning arch inspecting a friendly robot's glowing inner machinery.

The standard way to keep an AI agent in line is to have a second AI read over its shoulder. Goodfire, the interpretability startup, has just shipped an alternative: monitors that watch what is happening inside a model's calculations while it works, instead of auditing the words it produces afterward.

  • Read more about Goodfire's Inside-Out Monitors: Reading an Agent's Mind Beats Reading Its Transcript

Google's EmbeddingGemma 2: One Open Model for Text, Code, Images, and Audio

Cartoon illustration of a glowing magnet drawing documents, code brackets, photos, and sound waves into a single swirling vortex.

Buried under this week's frontier-model headlines, Google DeepMind shipped something many working developers will get more practical use out of than any trillion-parameter announcement: EmbeddingGemma 2, a 740-million-parameter open embedding model released under Apache 2.0. It maps text, code, images, video, and audio into a single shared vector space, carries an 8K-token context window, and ships with modular text, vision, and audio encoders designed for on-device deployment.

  • Read more about Google's EmbeddingGemma 2: One Open Model for Text, Code, Images, and Audio

Rivals at One Table: OpenAI, Anthropic, and DeepMind Coordinate on AI Safety

Cartoon illustration of three distinct robots shaking hands around a round table, one holding a shield, in a warm cooperative scene.

In a week full of model launches, the most unusual announcement involved no model at all. OpenAI, Anthropic, and Google DeepMind have confirmed they spent several weeks coordinating on AI safety, jointly unveiling cyber-focused AI models, safeguards, and access programs developed collaboratively across the three labs. One report describes a formally announced joint standards body publishing common testing protocols for frontier models.

  • Read more about Rivals at One Table: OpenAI, Anthropic, and DeepMind Coordinate on AI Safety

Pagination

  • First page « First
  • Previous page ‹ Previous
  • …
  • Page 5
  • Page 6
  • Page 7
  • Page 8
  • Page 9
  • Page 10
  • Current page 11
  • Page 12
  • Page 13
  • Next page Next ›
  • Last page Last »

Copyright © 2026 Rocky Mountain Madman LLC - All rights reserved

Developed and Designed by Rocky Mountain Madman LLC