Google published its official AI optimization guide in July 2026 and made one thing explicit: you do not need to create `llms.txt` files, special AI markup, or any new machine-readable files to appear in Google Search or its generative AI features. The guidance is direct — "Google Search ignores them."
Two weeks later, Shopify's developer documentation describes `/agents.md` as "the canonical agent-facing description of a store" and ships it by default on every Shopify storefront.
Both statements are accurate. They are not contradictory. But the gap between them is exactly where most businesses are getting confused — and where the next phase of web optimization is quietly taking shape.
Google's Position: Don't Invent a New SEO Layer
Google's AI optimization guide is the clearest official statement the company has made about what website owners actually need to do in the generative AI era. The short version: keep doing SEO.
The guide explains that Google's AI Overviews and AI Mode are built on the same core ranking and quality systems that power regular Search. They use retrieval-augmented generation (RAG) — the same scrape-and-ground pipeline that pulls from the Search index — and they reward the same signals: crawlability, indexability, unique non-commodity content, and accurate business information.
On the question of `llms.txt` and similar files, Google is unambiguous:
"You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them."
The guide goes further: creating and maintaining `llms.txt` files for other services is fine. Doing so will neither help nor hurt your Google Search visibility. Google simply ignores them.
The same applies to "chunking" content into smaller pieces for AI comprehension, rewriting content specifically for AI systems, and over-investing in structured data beyond what already supports rich results. None of these are required for Google's generative AI features.
What Google does want is straightforward: make your content crawlable, make it genuinely useful, keep your local business details accurate in Google Business Profile and Merchant Center, and ensure your pages load and display correctly. The guide explicitly frames this as "SEO is still SEO" — the fundamentals haven't changed, they've just become more important.
The one forward-looking note Google adds is about agentic experiences. The guide mentions browser agents that interact with websites by analyzing visual renderings, inspecting the DOM, and interpreting the accessibility tree. It points to the Universal Commerce Protocol (UCP) as an emerging standard. But it frames this as something to "explore if relevant to your business and you have extra time" — not a current requirement.
Shopify's Position: Agents Need Operational Instructions
Shopify is operating in a different context, and its `/agents.md` file reflects a different problem.
Shopify's developer documentation describes `/agents.md` as the canonical agent-facing description of a store. Its purpose is not to improve search rankings or help Google's AI Overviews cite the store more accurately. Its purpose is to help autonomous commerce agents understand what the store can do and how to interact with it.
The file exposes:
- What the store sells and how it's organized
- Which machine endpoints exist (UCP/MCP)
- How an agent should browse products
- What policies apply to agent-driven transactions
- What actions an agent may safely take and what it must not infer
Shopify also treats `/llms.txt` and `/llms-full.txt` as alternate URLs that mirror `/agents.md` by default — so all three paths can resolve to the same agent-facing information. The question Shopify is answering isn't "how do I rank in Google?" It's "how does an autonomous shopping agent understand and operate inside my store?"
That is a fundamentally different job.
The distinction matters because the agents Shopify is preparing for are not search crawlers. They are transactional agents — systems that may browse a product catalog, check inventory, apply a discount code, and complete a purchase on behalf of a user without any human clicking through a funnel. For those agents, a description of the store's capabilities and endpoints is genuinely useful. For Google Search's RAG pipeline, it's irrelevant.
The Files Are Not Competitors — They Answer Different Questions
The framing of `llms.txt` vs. `/agents.md` as competing standards misses the point. They serve different systems with different needs.
| File | Primary Consumer | Purpose |
|---|---|---|
| `llms.txt` | LLM-based research tools, RAG pipelines, AI assistants | Describe who the entity is, what it does, and what content is authoritative |
| `/agents.md` | Autonomous commerce agents, browser agents, UCP clients | Describe what the system can do, how to navigate it, and what actions are available |
| JSON-LD schema | Google Search, Knowledge Graph, AI Overviews | Provide structured entity facts for indexing and rich results |
| Visible prose | All of the above | Confirm entity facts for any system that strips HTML and reads plain text |
Shopify's own architecture reflects this. By defaulting all three paths (`/agents.md`, `/llms.txt`, `/llms-full.txt`) to the same content, Shopify is acknowledging that different systems may look for agent-facing information in different places. The file format is less important than the capability being described.
The better question for any website owner is not "which filename wins?" It's: what capability are you trying to expose, and which systems actually consume it?
For most businesses, the answer to the second question is still "Google Search" — which means the Google guidance applies directly. For businesses operating commerce infrastructure where autonomous agents are a real use case — large e-commerce platforms, multi-location service businesses with booking systems, healthcare networks with scheduling APIs — the Shopify model points toward something worth watching.
Search Crawlers and Browser Agents Have Different Needs
Google's guide touches on this distinction without fully resolving it. For Search, Google needs normal crawlable and indexable content. For autonomous interaction, an agent may additionally need to understand buttons, forms, labels, states, availability, routing, and transactional interfaces.
An `/agents.md` file can describe the environment. It cannot repair an unusable form or an ambiguous checkout workflow. The website still has to actually support the capabilities the file describes.
This is the grounding principle that keeps the conversation practical. Agent files are a description layer, not a fix. If your booking system is broken, your `/agents.md` file won't help an agent complete a booking. If your product inventory isn't accessible via a machine endpoint, listing a UCP endpoint in `/agents.md` doesn't create one.
The implication for non-Shopify sites is conservative: the infrastructure has to exist before the description file adds value. For most businesses, that infrastructure doesn't exist yet, and building it is a larger project than creating a text file.
What Non-Shopify Sites Should Actually Do
Google's guidance and Shopify's implementation together suggest a practical decision framework for website owners who aren't running a Shopify storefront.
Keep doing: Normal SEO fundamentals — crawlability, indexability, structured data for rich results, accurate Google Business Profile, useful non-commodity content. These are what Google's generative AI features actually use.
Do now: Make sure important business facts — locations, services, organizational relationships, hours, contact information — are present in plain, readable text on the relevant pages. Not buried in images. Not only in schema. In visible prose that any scraper can read. This is the entity synchronization work that makes the difference between being cited and being ignored.
Test if relevant: An `/agents.md` or `llms.txt` file if there is a credible agent use case and the maintenance cost is low. For a multi-location service business with a real booking API, a machine-readable description of how an agent should route requests is worth building. For a five-page brochure site, it's not.
Don't build: A sprawling "AI optimization" architecture because Shopify created one for its commerce platform. Shopify's infrastructure exists to support autonomous commerce agents at scale. That's a specific use case with specific technical requirements. Copying the file structure without the underlying capability is theater, not optimization.
The Closing Line That Actually Matters
Google is telling publishers how to remain useful to AI-powered search. Shopify is showing merchants how to become usable by AI-powered agents. Those are different problems, and the solutions are not interchangeable.
The businesses that will navigate this transition well are the ones that understand which problem they're actually solving. For most of them, the answer is still the same one it's always been: make your entity clear, make your content useful, and make sure the systems that matter can read both.
The agent layer is coming. But it builds on the foundation, not around it.
CTA
Your Entity Foundation Comes Before the Agent Layer
Before agent files, before llms.txt, before any of the emerging infrastructure — your entity facts need to be clear, synchronized, and machine-readable in the systems that matter today. That's the foundation the MFGSEO Enterprise Alignment Sprint builds.
Book an Alignment Sprint →About the Author
Alex Rodriguez is an AI-first SEO operator based in Cedar Park, TX. 15+ years building content systems that drive AI visibility and organic growth.
About Alex →Want This for Your Site?
I build content systems optimized for AI answer selection. Start with an audit.
Request an Audit