Understanding the Agentic Commerce Protocol
Why merchants must prepare for AI-driven shopping
AI shopping isn’t hype: it’s live.
Since September 29th, 2025, shoppers in the U.S. can discover and purchase products entirely inside ChatGPT , powered by OpenAI and Stripe . A request like:
“Find a hiking backpack under $200 with great reviews, deliverable to Colorado before Thanksgiving.”
…can now lead straight to checkout: no storefront, no tabs, no browsing.
For the first time, commerce can start and end inside a conversation.
That shift requires a new infrastructure standard: the Agentic Commerce Protocol (ACP) .
What is agentic commerce?
Agentic = handled by an AI agent on behalf of the shopper.
Instead of browsing pages, customers express their needs in natural language.
Agents interpret intent → fetch structured catalog data → complete the purchase.
The interface becomes intent and dialogue, not buttons and menus.
Why a protocol is needed
Webshops are designed for humans:
- Visuals and brand storytelling
- Creative layouts
- Paragraph-long descriptions
Agents can’t rely on that.
They require:
- Structured product data
- Real-time stock & pricing
- Clear attributes & variants
- Checkout flows they can automate
Without this data, merchants disappear from AI-driven discovery.
ACP ensures that machines can trust what you sell.
The scraping window is closing
Since ChatGPT came out, the web was crawlable at will.
Not anymore:
- July 2025 — Cloudflare rolls out AI Crawl Control (default deny)
- September 2025 — Google quitely disabled the num=100 search-parameter
- November 2025 — Amazon sues Perplexity over agentic shopping tool
The Web is saying no to industrial level AI scraping.
Scraping your human product page regularily is getting harder and more expansive for AI companies.
What ACP introduces
ACP standardizes how stores expose:
- Product data → what’s sold, without guesswork
- Availability & pricing → real-time data
- Variants & attributes → clear sizes, models, materials, etc
- Trusted checkout → merchant remains merchant of record
It’s a connective layer between ecommerce systems and AI agents.
Why this matters now
For twenty years, ecommerce visibility required:
- being ranked by a search engine
- looking good in a browser
But in agentic commerce, machines interpret and surface products by relevance.
If your product data isn’t exhaustive, up-to-date and optimized for intent, it won’t appear in the conversation.
Some brands will silently vanish from the conversation.
Who’s behind ACP
This isn’t a niche side-project.
Companies driving and adopting ACP include:
| Company | Role |
|---|---|
| OpenAI | Conversational search + checkout inside ChatGPT |
| Stripe | Commerce trust + payments infrastructure |
| Etsy | First marketplace live with instant checkout |
| Shopify | Preparing ACP adoption across ~1M merchants |
| Walmart | AI-powered product matching and ordering |
| PayPal | Wallet/payment network preparing ACP support |
ACP emerges from the combined force of leading commerce platforms, payment networks, and AI agents.
Such a coordinated shift signals a generational change in how people shop.
The inflection point
Complex buying journeys still need discovery, comparison, exploration.
But the many practical, simple, low-emotion purchases now have a better path with agent-driven commerce.
On top of that, your favorite LLM knows you better than Google.
You just say:
I need a protein bar.
The agent already knows that you’re vegetarian, allergic to nuts, training, prefer chocolate.
No need to scroll 10 websites: the AI does the matching.
Those transactions will leave webshops and never come back : for some merchants, that’s their entire catalog.
But even if some buyers won’t transact in the conversation, for most, the product discovery will start here.
That’s the real shift.
Looking ahead
We are early.
Today, ACP is only released in the US.
Europe in 2026.
OpenAI did not yet release any official numbers for Instant Checkout or ACP commerce.
Good or bad at first, numbers will grow until conversations become major ecommerce channels: McKinsey projects agent-driven commerce to generate up to $1 trillion in U.S. retail revenue by 2030 and $3-5 trillion globally.
Now is the time for merchants to:
- adapt to be interpreted, not browsed
- prepare their catalogs to become machine-ready
You can now see how ready your Shopify product pages are for AI agents .
If you’re also exploring ACP or agent-driven commerce, let’s connect on LinkedIn :)