AI discovery is becoming more fragmented.
ChatGPT, Claude, Perplexity, Google's AI experiences, and emerging shopping assistants do not all retrieve information in the same way. They use different systems, different crawlers and user agents, and different methods for deciding what information to surface.
For merchants, that creates a practical problem:
Should you optimize your product catalog separately for every AI platform?
Probably not.
A more durable approach is to make the underlying product information accessible, structured, and unambiguous enough that different AI systems can work with the same source of product truth.
That is the infrastructure problem Agentic Page is designed to address.
TL;DR
- AI discovery is multi-platform. No single AI assistant or crawler is guaranteed to remain the dominant path through which shoppers discover products.
- Agentic Page is platform-agnostic by design. It organizes a merchant's existing product information into a structured, AI-readable surface rather than creating a separate fix for each AI platform.
- Readability is infrastructure, not a recommendation guarantee. Making product information accessible and clear gives AI systems better material to work with. It does not determine whether a specific platform will recommend, cite, rank, or convert a product.
Why optimizing for one AI platform is a moving target
The AI landscape is changing too quickly for merchants to build their entire discovery strategy around whichever platform appears most important today.
A platform that sends meaningful traffic this quarter may become less important as another assistant, shopping interface, or AI-native experience gains adoption.
That means platform-specific optimization can become expensive to maintain.
A merchant might spend months tailoring content to the requirements or quirks of one system, only to find that a different AI surface becomes more relevant later.
The more durable question is not: “How do we optimize for one AI platform?”
It is: “Can our product information be accessed and understood reliably across AI systems?”
That shifts the problem from platform optimization to product infrastructure.
What does “AI-readable” actually mean?
AI systems do not experience a product page the way a shopper does.
A human can scan a layout, interpret an image, click a size guide, open an accordion, compare several tabs, or contact customer support when information is missing.
A crawler or automated retrieval system works differently. It typically relies on machine-accessible elements such as:
- HTML and page text
- Structured product data
- Crawl permissions
- Clear product attributes
- Variant information
- Pricing and availability context
- Shipping and return information
This creates three basic requirements.
1. The page has to be accessible
Crawler access can be affected by robots.txt rules, CDN configurations, WAF settings, server behavior, or other technical controls. A product page that cannot be reached cannot contribute useful information to that retrieval path.
2. The important information has to be machine-readable
Critical product facts should not exist only inside images, interactive elements, or client-side experiences that are difficult for automated systems to interpret. A shopper may understand a visual size chart immediately. A machine may need the same information represented clearly in text or structured data.
3. Product facts have to be unambiguous
Being crawlable is not the same as being understandable. An AI system still needs to determine:
- What exactly is the product?
- Which variants exist?
- Which attributes belong to which variant?
- What is included?
- What is the current price?
- Is it available?
- What are the shipping and return conditions?
- Which use cases is it designed for?
The less guessing required, the stronger the underlying product information becomes for AI-mediated discovery.
Why merchants should not build around a fixed crawler list
Today, a Shopify store may encounter crawlers and AI-related user agents associated with companies such as OpenAI, Anthropic, Perplexity, Google, and other emerging AI platforms.
That list will continue to change.
New AI shopping experiences will appear. Existing platforms will change how they retrieve information. Regional and vertical AI assistants may become important for specific categories.
For merchants, the goal should therefore not be to maintain a permanent checklist of five crawler names.
The goal should be to maintain product infrastructure that remains accessible and interpretable as the ecosystem evolves.
Crawler-specific checks still matter, especially when diagnosing access problems. But the underlying product truth should remain platform-independent.
One product truth, many AI surfaces
This is where Agentic Page takes a different approach.
Instead of creating a separate content layer for each AI platform, Agentic Page organizes a merchant's existing product and brand information into a structured, AI-readable surface. That includes information such as:
- Product identity and descriptions
- Variants and attributes
- Pricing and availability
- Use cases
- Shipping information
- Return policies
- Other relevant product facts
The same underlying product truth can then be made available to different compliant AI crawlers and retrieval systems.
When the Shopify catalog changes, that information can remain synchronized rather than relying on a collection of static, platform-specific patches.
This matters because channel behavior may fragment, but product truth should not. A merchant should not need five different versions of what a product is simply because five AI systems may read it differently.
Does this mean every AI platform will treat the product the same way?
No. This distinction is important.
Making product information accessible, structured, and unambiguous is a prerequisite for accurate machine understanding. It is not a guarantee that every AI platform will:
- Retrieve the product
- Rank it highly
- Cite it
- Recommend it
- Send a visit
- Generate an order
Different AI systems use different retrieval methods, ranking logic, trust signals, context, and recommendation systems. Agentic Page does not control those decisions.
What it addresses is a more fundamental and controllable failure point: a product should not be overlooked or misrepresented simply because its underlying information was difficult for an AI system to access or interpret.
That is why AI readability should be understood as infrastructure rather than a ranking promise.
What should Shopify merchants do now?
A practical AI-readiness strategy can start with four steps.
- Check crawler accessibility regularly. Crawler access is not a one-time technical task. A site migration, new CDN rule, firewall configuration, theme change, or infrastructure update can create new access problems even after previous issues were fixed. Accessibility should be monitored over time.
- Build for product truth, not one platform. Avoid making one AI platform the foundation of your entire AI-discovery strategy. The ecosystem is still changing too quickly. Instead, keep the underlying product information accurate, structured, and reusable across different AI surfaces.
- Keep the catalog as the source of truth. Platform-specific fixes become difficult to maintain when they drift away from the merchant's actual catalog. Product information should stay synchronized with the core store so that changes in variants, pricing, availability, policies, and product details do not create conflicting versions of the same product.
- Measure readability and business outcomes separately. This is one of the most important distinctions in AI commerce. A crawler successfully accessing a product page is one measurement. That product appearing in an AI answer is another. An AI-referred visit is another. A conversion or attributed order is another. They belong to the same funnel, but they are not the same metric. Merchants should measure each stage separately rather than treating crawler activity as proof of recommendation or revenue.
Where Agentic Page fits
Agentic Page is DeepLumen's infrastructure layer for making Shopify product information easier for AI systems to access and understand. It transforms existing merchant data into an AI-readable product surface and keeps that layer aligned with the underlying catalog.
Across DeepLumen's network, more than 500,000 Agentic Pages have now been deployed, while cumulative AI crawler visits have surpassed 1.1 million. Those milestones show growing machine access to AI-readable product infrastructure. They should not be interpreted as a guarantee that every crawl produces a recommendation, visit, or order.
The role of Agentic Page is earlier in the journey: make product truth accessible, structured, current, and easier for AI systems to interpret. From there, each AI platform still decides what it retrieves, how it evaluates products, and what it ultimately presents to the shopper.
That distinction is fundamental to how DeepLumen thinks about Agentic Commerce.
AI visibility begins with being readable. Recommendation comes later.
Build for the ecosystem, not one crawler
The AI shopping ecosystem will continue to change. New assistants will emerge. Retrieval systems will evolve. Shopper behavior will shift between platforms.
Trying to predict one permanent winner is difficult. Building product infrastructure that remains useful across those changes is much more durable.
For Shopify merchants, the objective is not to optimize for every AI system one by one. It is to make sure that when an AI system comes looking for product information, there is something accurate, structured, and current for it to find.
That is the role of the AI-readable discovery layer.
Related reading: How Agentic Page Creates an AI-Readable Discovery Layer · Why AI Shopping Channels Don't Share the Same Economics · How AI Answers Are Becoming a New Discovery Channel
Want to understand how readable your product catalog is to AI systems today? Learn more about Agentic Page or book a demo to review your store's AI discovery readiness.