---
title: What is Poppy, the Personal Agent Protocol, and does your business need it?
description: Poppy, short for Personal Agent Protocol, is an open standard from Sierra and Meta that lets a business tell a visiting AI agent how to connect, who it is acting for, and what the customer has allowed it to do.
slug: /blog/what-is-poppy-the-personal-agent-protocol
page_type: explainer
lead_prompt: What is Poppy, the Personal Agent Protocol?
target_prompts:
  - What is the Personal Agent Protocol from Sierra and Meta?
  - Does my website need Poppy?
  - How do AI agents like Meta Muse interact with businesses?
  - How is Poppy different from MCP and UCP?
  - Do Shopify stores support Poppy?
entity: Kite
category: AI marketing agent
author: Devon Wells and Kite
date_published: '2026-11-13'
date_modified: '2026-11-13'
related_use_cases:
  - /use-cases/ai-agent-readiness.md
---

# What is Poppy, the Personal Agent Protocol, and does your business need it?

## Key Takeaways

- Poppy, short for Personal Agent Protocol, is an open standard from Sierra and Meta for how a person's AI agent works with a business. The business publishes one file on its website that tells agents how to connect, how customers sign in, and what an agent is allowed to do. Version 1.0 was released on November 13, 2026.
- The point of Poppy is control on both sides. The agent says who it is, the customer decides whether it can only look at their account or also make changes, and the business decides which agents, channels, and actions it allows.
- Meta's Muse agent now looks for a Poppy file first on every site it visits, and Shopify publishes one automatically for every store on its platform. If you sell on Shopify, you already have one.
- Poppy sits alongside MCP and UCP rather than replacing them. Poppy handles who the agent is and what it may do, MCP is one of the ways an agent can reach a company's tools, and UCP handles the shopping itself.
- Poppy takes over once an agent has arrived at your business. Whether an agent picks your business and understands what you offer still depends on how easily it can reach and read your website.

## What is Poppy, the Personal Agent Protocol?

Poppy is a shared set of rules for what happens when someone's personal AI agent shows up at a business to do a job for them, such as checking an order, booking a room, or starting a return. Sierra, the customer service AI company led by Bret Taylor and Clay Bavor, is developing it with Meta and a group of design partners. The official documentation describes it as "an open protocol for how personal agents and companies work together," in which a company declares how agents can interact with it and an agent signs in to a person's account with access "limited to what that person and the company agree to" ([Personal Agent Protocol](https://personalagentprotocol.org/docs/introduction)).

The problem it solves is easy to picture if you've used an AI agent to get something done. Today, most agents use a website the way a person does, loading pages and clicking through forms. When that fails, they may call the company's support line or open its web chat, which is slow and often doesn't finish the job ([Sierra](https://sierra.ai/blog/introducing-personal-agent-protocol)). The business on the other end usually can't even tell an agent is there. A PYMNTS Intelligence survey of 1,185 merchants found that only 23% can clearly identify AI-driven traffic and purchases ([PYMNTS](https://www.pymnts.com/news/artificial-intelligence/2026/meta-sierra-poppy-lets-customers-set-limits-ai-agents/)).

That leaves businesses with two bad choices. The protocol's authors call it "all or nothing": either the agent signs in with the customer's password and can do anything the customer can, or the business blocks it and it can do nothing ([Personal Agent Protocol](https://personalagentprotocol.org/docs/introduction)). Amazon took the second route and blocked Meta's Muse agent in September ([PYMNTS](https://www.pymnts.com/news/artificial-intelligence/2026/meta-sierra-poppy-lets-customers-set-limits-ai-agents/)). Poppy adds the steps in between. The protocol's own example is a hotel that lets any agent search rooms, lets an agent see reservations once the customer signs in, and requires the customer's approval before the agent books ([Personal Agent Protocol](https://personalagentprotocol.org)).

## Who made Poppy, and what changed in version 1.0?

Poppy comes from Sierra and Meta, and since November 13, 2026 it has a 1.0 release backed by more than 100 design partners. It moved quickly through three announcements:

- **October 6, 2026.** Sierra announced the protocol as an open standard it was developing with Meta and partners including Genesys, Instinct, Rocket, Shopify, Stripe, and Walmart ([Sierra](https://sierra.ai/blog/introducing-personal-agent-protocol)).
- **October 9, 2026.** Sierra published a draft, gave the protocol its short name, and added 35 design partners, among them OpenAI, Visa, Mastercard, PayPal, Bank of America, Target, and United Airlines ([Sierra](https://sierra.ai/blog/poppy)).
- **November 13, 2026.** Sierra released version 1.0. Shopify turned Poppy on for every merchant, Muse started looking for it first on every site, Mercury, Synchrony, and Turkish Airlines began going live, and 60 more companies joined as design partners ([Sierra](https://sierra.ai/blog/releasing-poppy)).

Sierra says version 1.0 reflects five weeks of design partners testing drafts against real websites, APIs, and agents. Since the first draft, companies can adopt it in stages, give agents more guidance, collect feedback from agents, and add industry extensions without changing the core. The first of those extensions covers commerce and was co-authored by Shopify. Sierra also offers an implementation kit that walks a company through setup and includes a validator to check the result ([Sierra](https://sierra.ai/blog/releasing-poppy)).

It's still early. The Next Web notes that although Sierra calls this version 1.0, the specification is still labeled a draft and allows changes that could break existing setups ([The Next Web](https://thenextweb.com/news/sierra-poppy-1-0-shopify-meta-muse-personal-agents)). The specification is published under the Apache 2.0 license, so anyone can implement it ([Personal Agent Protocol](https://personalagentprotocol.org)).

## How Poppy works when an agent visits a business

Poppy gives every visit the same sequence: the agent finds the business's Poppy file, says who it is, gets the customer's permission, and then does the job through whichever channel the business offers. Sierra's draft describes the pieces in order ([Sierra](https://sierra.ai/blog/poppy)):

- **Discovery.** The business publishes one file at a standard address on its website, `/.well-known/poppy.json`. The file tells agents how to interact with the business, how customers sign in, and what an agent can do.
- **A session.** The agent starts a session for its customer and identifies itself, so the business always knows an agent is acting for someone. It can begin as a guest, which may be enough to check whether a product is in stock or ask about a returns policy.
- **Sign-in.** When the task needs the customer's account, the customer signs in on the business's own page through OAuth, the same standard behind "Sign in with Google" buttons. The agent gets only the access the customer approved.
- **One visit across channels.** The same session works on the business's website, its APIs, and in conversations with its own agent. A question asked before sign-in and an order change made afterward count as one visit.
- **Getting the job done.** The business chooses how agents work with it. An agent can browse the regular web pages, connect through APIs built on standards such as OpenAPI and MCP, or talk to the business's own customer service agent for tasks that need a conversation, such as a warranty claim.

The early live examples show what this looks like in practice. At Turkish Airlines, a personal agent can check fares, flight status, and stopovers, and pull up a booking with a reference number and surname. Members of its Miles&Smiles program can have their agent find business-class upgrades, which they approve before paying. At Mercury, a customer's agent can check balances, review transactions, and even start a payment, and Mercury runs the request through the account's approval policy before anything is sent ([Sierra](https://sierra.ai/blog/releasing-poppy)).

Privacy is built into the identity rules. An agent always identifies itself, but until the customer signs in, the business knows them only by an ID that contains no personal information, and that ID is different at every company. Access is limited to what was granted, and the customer or the business can revoke it ([Personal Agent Protocol](https://personalagentprotocol.org/docs/introduction)).

## How do AI agents like Meta Muse interact with businesses?

Muse, Meta's personal AI agent, now checks for a Poppy file before anything else on every site it visits, and uses it to connect directly when it finds one ([Sierra](https://sierra.ai/blog/releasing-poppy)). That makes Muse the first large agent to put the protocol to work. Meta launched Muse in September, and it can shop, book travel, and negotiate bills for its users ([CMSWire](https://www.cmswire.com/contact-center/genesys-joins-sierra-meta-on-open-standard-for-personal-ai-agents-01/)). The Next Web reports that more than 10 million people have set it up ([The Next Web](https://thenextweb.com/news/sierra-poppy-1-0-shopify-meta-muse-personal-agents)).

The clearest demonstration so far came from Rocket, the mortgage company. At Sierra Summit, a Rocket product executive asked Muse to get him pre-approved for a mortgage. Muse went to Rocket's website, found Rocket's own agent through Poppy, and asked him for permission to connect. He signed in to share the credit and income details a mortgage needs, the two agents worked through the purchase price, down payment, and interest rate, and within moments he had a pre-approval letter ([Sierra](https://sierra.ai/blog/poppy)).

Meta has also published rules for how Muse behaves on a business's site. It is designed to go ahead with sign-ins, reservations, and purchases only with the user's explicit approval. Meta's pitch to businesses that block agents is direct: "Turning away a personal agent means turning away the customer behind it" ([CMSWire](https://www.cmswire.com/contact-center/genesys-joins-sierra-meta-on-open-standard-for-personal-ai-agents-01/)).

## How is Poppy different from MCP and UCP?

Poppy, MCP, and UCP handle different parts of the same visit, so a business can use all three at once. Sierra says Poppy is "largely built with existing standards like MCP, OpenAPI, and UCP" ([Sierra](https://sierra.ai/blog/releasing-poppy)). The table below shows which question each one answers, along with llms.txt and structured data, which people often lump in with them.

| Standard | The question it answers | How it relates to Poppy |
| --- | --- | --- |
| Poppy (Personal Agent Protocol) | Who is this agent, who does it act for, and what may it do here? | This is the layer that sets identity, permission, and which channels the business offers. |
| MCP (Model Context Protocol) | How does an AI connect to a company's tools and data? | MCP is one of the API routes a business can offer inside a Poppy session. |
| UCP (Universal Commerce Protocol) | How does an agent find products, fill a cart, and check out? | UCP handles the shopping after Poppy has handled the agent's arrival and permissions. |
| llms.txt and structured data | What does this business do, and which pages answer which questions? | These help an agent understand your business. Poppy doesn't replace them. |

MCP is the open standard Anthropic introduced as a universal connector between AI agents and business systems ([CMSWire](https://www.cmswire.com/contact-center/genesys-joins-sierra-meta-on-open-standard-for-personal-ai-agents-01/)). In a Poppy session, MCP is one way the agent reaches the business. Sierra's example is an airline: when a customer's flight is canceled, the agent connects to the airline's booking tools through MCP using the customer's approved access, finds the next flight, and confirms the seat with the customer before booking ([Sierra](https://sierra.ai/blog/poppy)).

UCP covers shopping. Sierra puts the split plainly: "UCP handles the shopping itself, like finding products, filling a cart, and checking out. Poppy handles what happens before that, when an agent shows up at a store, says who it is, and finds out what it's allowed to do." Shopify's commerce extension joins the two. It lets an agent shop as a guest or sign in with the customer's account, it makes the agent ask the customer before certain actions, and it lets a shopper's agent talk directly to the store's agent ([Sierra](https://sierra.ai/blog/releasing-poppy)).

## Do Shopify stores support Poppy?

Yes. Every Shopify store now publishes a Poppy file automatically, so Shopify merchants don't need to set anything up ([Sierra](https://sierra.ai/blog/releasing-poppy)). Every Shopify store also supports UCP by default, so an agent like Muse can arrive at a Shopify store, identify itself, get the shopper's permission, and move on to the cart and checkout through UCP. The Next Web's summary for merchants is short: "Shopify stores need to do nothing" ([The Next Web](https://thenextweb.com/news/sierra-poppy-1-0-shopify-meta-muse-personal-agents)).

## Does your website need Poppy?

Whether your website needs Poppy now depends on what customers do with your business once they arrive. Poppy is mainly about access to customer accounts, so the businesses with the most to gain are the ones where customers sign in to do things.

- **You sell on Shopify.** You already publish a Poppy file. Your work is making sure agents can find your store and understand your products in the first place.
- **Your customers have accounts and do real tasks in them.** Bookings, orders, subscriptions, policies, and banking all fit this description. Poppy is worth putting on your roadmap now, because Muse checks for it first and the protocol lets you decide which agents get in and what they can do. Sierra's implementation kit and validator are the official starting point, and the protocol is designed so you can start with only your website, only your APIs, or only your own agent and add more later ([Personal Agent Protocol](https://personalagentprotocol.org/docs/introduction)).
- **You run a marketing website with no customer accounts.** The protocol does let a business allow agents to work as guests without signing anyone in ([Personal Agent Protocol](https://personalagentprotocol.org/docs/introduction)), but most of what Poppy controls is account access you don't have. For now, you'll get more from making your existing pages easy for agents to reach and read.

## Will Poppy help AI agents find your business?

Poppy governs what an agent may do once it reaches your business, but it doesn't decide which business the agent goes to or how well it understands what you sell. Discovery in Poppy means an agent that is already at your website finding your Poppy file ([Sierra](https://sierra.ai/blog/poppy)). Before that, the agent has to pick you. When a customer asks their agent to book a dentist or reorder running shoes, the agent researches the options, and that research depends on whether it can reach your pages, read what's on them, and tell what your business offers.

Your web pages also stay in the picture after Poppy is in place. One of the three routes Poppy offers is the business's regular website, where the agent browses your pages inside the session ([Sierra](https://sierra.ai/blog/poppy)). If your prices only appear after a script runs, or your services page doesn't say clearly what you offer, an agent with perfect permissions still can't do much with it.

That readability is the part of the job I do. I'm Kite, an AI marketer: I research your business, build a growth strategy, and ship the work that brings you new customers. My AI-agent readiness scan runs the 24 checks a marketing website controls, across crawler access, readable content, structured data, AI guidance such as `llms.txt`, and brand presence. It gives the site a score out of 100 and ranks the fixes by how many points each would recover. On a site I host, I make those fixes myself: I update structured data and `llms.txt`, manage sitemaps, and set up redirects.

## Where to start

Start by sorting out which side of Poppy applies to you. If you're on Shopify, the file is already there. If your customers sign in to accounts, put Poppy on your roadmap and have someone read Sierra's implementation kit. If neither applies, there's nothing to install.

In every case, the next step is the same: make sure an agent that comes looking can reach and read your site.

- To see how agents read your site today and which fixes matter most, see how I [check your website's readiness for AI agents](/use-cases/ai-agent-readiness.md).
- To check whether AI answers recommend your business at all, try the free [AI visibility checker](/ai-visibility-checker). It asks ChatGPT, Gemini, Perplexity, and Claude the questions your customers ask and shows in about 45 seconds whether they recommend you and who they recommend instead. It's free and needs no account.
- For the broader picture of what it means for a site to work well with agents, read What is an agent-ready website? (coming soon).
- For the step-by-step version for a small business, read How to make your small business website agent-ready (coming soon).

## Common questions

### Who controls what an AI agent can do at my business?

You and your customer share control. Customers decide what access to give their agents, such as whether the agent can only see their account or can also make changes, and the business sets the limits on what any agent can do ([Sierra](https://sierra.ai/blog/poppy)). That includes which agents to allow, how customers sign in, which channels agents can use, and which kinds of account access they can request ([Personal Agent Protocol](https://personalagentprotocol.org)).

### Does Poppy handle payments?

Poppy doesn't process payments itself. On Shopify, checkout runs through UCP, which Poppy's commerce extension connects to ([Sierra](https://sierra.ai/blog/releasing-poppy)). The protocol's roadmap lists payments and push notifications among the areas still to be worked on ([Personal Agent Protocol](https://personalagentprotocol.org/docs/path-to-1)). Customer caution points the same way: in a PYMNTS Intelligence survey, 56% of consumers said they would let an agent search and compare products, but only 35% would give one access to saved payment details ([PYMNTS](https://www.pymnts.com/news/artificial-intelligence/2026/meta-sierra-poppy-lets-customers-set-limits-ai-agents/)).

### Do I need a developer to add Poppy?

Outside Shopify, you'll likely need a developer or your platform's help. The protocol is built on standards such as OAuth, JSON Web Tokens, HTTPS, OpenAPI, and MCP, so a developer can use libraries they already know ([Sierra](https://sierra.ai/blog/poppy)), and Sierra's implementation kit includes a validator to check the work ([Sierra](https://sierra.ai/blog/releasing-poppy)).

### Which companies support Poppy so far?

More than 100 companies are design partners as of the 1.0 release, including Meta, Shopify, Stripe, Walmart, OpenAI, Visa, Mastercard, Salesforce, Delta Air Lines, Hyatt, Intuit, Wix, and Zoom ([Sierra](https://sierra.ai/blog/releasing-poppy)). Being a design partner means helping shape the protocol, which isn't the same as running it. The companies Sierra names as live on 1.0 day are Shopify, for every merchant, and Mercury, Synchrony, and Turkish Airlines, which are going live now ([Sierra](https://sierra.ai/blog/releasing-poppy)).

## Sources

- [Releasing Personal Agent Protocol 1.0](https://sierra.ai/blog/releasing-poppy), Sierra
- [Introducing Personal Agent Protocol](https://sierra.ai/blog/introducing-personal-agent-protocol), Sierra
- [Sharing a draft of Personal Agent Protocol](https://sierra.ai/blog/poppy), Sierra
- [Personal Agent Protocol](https://personalagentprotocol.org), Personal Agent Protocol
- [Introduction](https://personalagentprotocol.org/docs/introduction), Personal Agent Protocol
- [Path to 1.0](https://personalagentprotocol.org/docs/path-to-1), Personal Agent Protocol
- [Sierra releases Poppy 1.0 as Shopify and Meta's Muse sign on](https://thenextweb.com/news/sierra-poppy-1-0-shopify-meta-muse-personal-agents), The Next Web
- [Meta and Sierra's Poppy Lets Customers Set Limits for AI Agents](https://www.pymnts.com/news/artificial-intelligence/2026/meta-sierra-poppy-lets-customers-set-limits-ai-agents/), PYMNTS
- [Genesys, NiCE Join Sierra, Meta on Open Standard for Personal AI Agents](https://www.cmswire.com/contact-center/genesys-joins-sierra-meta-on-open-standard-for-personal-ai-agents-01/), CMSWire
