Discussion starter: an AI search question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- AI Search
Should brands publish content mainly for AI agents to consume, like structured specs, stock and policies, even if few humans read it?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
Niraj Raut
Publish structured, factual information that agents can use, but publish it for people too, in the same places, and supply it through the channels agents actually read. Clear specifications, stock, prices, delivery and returns policies, compatibility and eligibility rules help human shoppers as much as software, so there is no need for content that “few humans read”. Google’s May 2026 guide to optimising for its generative AI features says site owners do not need special AI text files, Markdown versions or AI-specific formats for its features. Where agents do use dedicated channels, they are mostly feeds and commerce protocols: Merchant Center product data, Google’s commerce protocol for checkout in its AI features, and similar programmes from other platforms. Hidden or separate content aimed only at agents risks inconsistency and, at worst, looks like cloaking.
What agents need, and where it should live
Information On the page In feeds or protocols Specifications and compatibility HTML tables with clear labels and units Product attributes in feeds Price and availability Visible, current, including variants Feeds updated as often as prices change Delivery and returns policies Clear policy pages and summaries on product pages Merchant Center settings and structured data Eligibility and restrictions Plain statements of who can buy and where Feed attributes where supported Answers to common pre-purchase questions On product pages Conversational attributes in Merchant Center, where available Channels agents actually use
- Product feeds. Merchant Center data feeds Google’s shopping surfaces and AI shopping features. Merchant Center has added conversational attributes, available globally, aimed at the questions shoppers ask.
- Commerce protocols. Google’s Universal Commerce Protocol lets shoppers check out in AI Mode and Gemini with participating merchants; as of September 2026 it covers selected merchants and products in the US, Canada and Australia, with eligible US Shopify merchants included automatically since 22 September. OpenAI’s Agentic Commerce Protocol continues for its platform.
- The page itself. Agentic browsers read and operate pages, relying on clear structure and accessible controls.
What I would not do
- Create hidden pages or text visible only to crawlers or agents.
- Maintain a separate “agent version” of facts that can drift from the human version.
- Invest heavily in formats Google says it does not need, expecting search benefit.
Some developer tools and agents do read formats like llms.txt or Markdown versions of documentation; providing them for those audiences is reasonable if they are generated from the same source as the pages and kept consistent.
Hypothetical example:
An electronics retailer considers building a separate machine-readable catalogue for AI agents. Instead, it rebuilds product pages with complete specification tables, compatibility lists and plain delivery and returns summaries, enriches its Merchant Center feed with the same attributes and conversational answers, and joins its region’s AI checkout programme where eligible. Human conversion improves because questions are answered on the page, customer service contacts fall, and the retailer’s products begin appearing in AI shopping answers with accurate details. There is one source of truth, used by people, search engines and agents alike.
Watch for this:
Inconsistent facts between the page, the feed and any agent-facing format are the biggest risk. An agent that reads one price and a page that shows another creates failed transactions and complaints. Generate every format from the same data, and check for mismatches regularly.
Policies deserve the most attention
Product details are usually well covered; policies often are not. Agents and shoppers both need to know delivery times by region, costs and thresholds, returns windows and conditions, warranty terms and any restrictions, stated precisely. Many sites bury these in long legal pages. A short, plain summary on product pages and a clear policy page, kept consistent with Merchant Center settings and structured data, answer the questions that most often stall purchases, whether a person or an agent is doing the buying.
Where the value is heading
As agents complete more tasks, the accuracy and completeness of product and policy data become a competitive factor: an agent choosing between retailers will favour the one whose information is clear and reliable. That is a strong reason to invest in data quality and feeds, and a weak reason to create content for agents alone.
Publish structured facts on pages people use, supply the same data through feeds and commerce protocols where platforms read them, avoid separate or hidden agent-only content, and keep every format consistent from one source of truth.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.