Frequently asked questions

How do AI agents actually understand my website?

Agents pull different signals depending on how they are built. Some take a screenshot of the rendered page and reason about it visually. Most assistants (ChatGPT browse mode, Claude's web tool, Perplexity, Gemini deep research) primarily read the raw HTML. Browser agents that point-and-click also lean on the accessibility tree. That is the same structural representation screen readers use. Marie Haynes, in her Lighthouse Agentic Browsing walkthrough, says that tree "tells an agent where the buttons are and what the important elements are."

AgentReadyWebsite installs files that strengthen the HTML and discovery channel: llms.txt, agenticweb.md, llm-info.html, JSON-LD, agent-permissions.json, and the rest of the files we ship. Most assistants reach for that machine-readable summary first when they need to learn your business, which is more efficient than reconstructing it from noisy HTML.

Accessibility-tree quality comes from your existing HTML semantics: landmarks, headings, alt text. We surface accessibility-tree integrity as an "Other gaps we identified" item in the Pre-Check rather than folding it into the score. This MVP leaves your existing HTML in place and installs agent-readiness files beside it.

Source: Marie Haynes, How to use the new Lighthouse Report to see if your website is agent ready (May 2026).

How many files do I get in my agent-readiness package?

Typically 13 files; the package is sized by what your site actually needs. Every order includes the standards-aligned core files, and sitemap.xml is generated only when your site does not already have one. Sites with commerce capability or a public REST API receive populated .well-known/ucp and .well-known/api-catalog files instead of placeholders.

The standards behind each file matter more than the count. Every file maps to a published standard, working-group proposal, or documented convention. See the full file list for what each one does and who governs it.

What is the Agent Readiness Score?

Your Agent Readiness Score runs from 0 to 100 and carries a letter grade from A to F. You can get your Agent Readiness Score in seconds from the free Pre-Check. The shareable snapshot uses the same Agent Readiness Score we show on the Agent Readiness Dashboard.

We build it from four questions. Each question looks at files we install or upsert. A high score means those files are present and well-shaped.

• Can an agent find you? Discoverability
  • robots rules for AI crawlers
  • sitemap
  • llms.txt
  • the agent card
• Can it understand your business? Content Understanding
  • structured data (schema.jsonld)
  • the AI information hub (llm-info.html)
  • agenticweb.md
• Can it use tools you publish? Agentic Capabilities
  • skills index
  • MCP server card
  • API catalog
  • WebMCP registration
• What are agents allowed to do? Permissions and Governance
  • agent permissions
  • commerce terms
  • Content-Signals

Citation in ChatGPT or Perplexity depends on the query and the competition.

The Pre-Check also warns when a host firewall returns 403 to automated readers before they can read robots.txt. That warning sits outside the score. Each Pre-Check run also saves a shareable public score snapshot.

A free scanner already gave me an agent-readiness score. How is this different?

Free agent-readiness scanners, including Cloudflare’s free scanner, confirm that agent-readiness matters. Their published data also shows almost nobody has these files yet. A scanner still ends with a to-do list. Most of them assume you already have a coding agent and a developer workflow to finish the work.

Some free scanners lean on server-side checks like Markdown content negotiation, and they may score llms.txt differently than Google Chrome Lighthouse does. We still ship llms.txt as foundational.

AgentReadyWebsite analyzes your actual site and generates the standards-aligned package for it. We deliver it ready to install. Domain Watch then checks that those files stay on your site for 12 months. You can run a free scan before and after if you want. Many of the files those scanners check for are the files we install.

Do you change my website's code?

No. AgentReadyWebsite delivers static files and installation guidance. We do not modify your theme, existing plugins, or server code, create live server-side endpoints, host your website, or change how your pages render. For WordPress websites, our free installer plugin installs your agent-readiness package from WordPress admin after you upload the delivery zip. No FTP is required. After install, the plugin also prints the package schema.jsonld in the public page head so HTML-only scanners see the same structured data as the file URL. On other platforms, upload the zip to your website root and .well-known folder. The install typically takes about 5-10 minutes. We can host an optional call with you to walk you through the installation.

Do you support WordPress, Shopify, Squarespace, Framer, Wix, or Webflow?

The free Pre-Check detects your CMS platform and shows the install path before checkout, so you see the path before you pay.

• WordPress: yes. Install with our free plugin from WordPress admin (upload your delivery zip; no FTP).
• Webflow, Framer, and most static or self-hosted sites: yes. Upload the package zip using our platform-specific instructions (typically about 5–10 minutes).
• Shopify, Squarespace, and Wix: coming soon. Pre-Check holds your spot with early access and we will notify you when they are ready.

What does the package include, and how does billing work?

You pay once. Basic is $199 for up to 50 pages, Medium $599 for 51 to 500 pages, and Complex $1,399 for 501 or more. The zip and 12 months of Domain Watch are included. We email a secure download link when your package is ready, usually within a few minutes. Download links expire in about 7 days. We keep the zip available for the watch year. Checkout is Stripe. See pricing.

What we do after you pay is on What happens after I pay?

What happens after I pay?

We generate the package and email the zip. You install the files. For 12 months we check that those files stay on your site. We email you every month, and sooner if a file is missing or robots.txt starts blocking a named AI bot.

If a services, pricing, FAQ, or about page changes in substance, we can offer a new generation. We have not set that price. See 12-month Domain Watch.

What do you deliberately not do?

We deliver a technical foundation so your website is more machine-readable and easier for LLMs and AI agents to understand. We leave website ownership with you. We leave traffic outcomes and Google AI Overviews or AI Mode placement unpromised. Specifically:

• No full-service SEO retainer
• No theme edits or custom development required. For WordPress websites, install via our free installer plugin (included with your purchase)
• No hosting, mirroring, or bidirectional sync
• No guaranteed traffic or revenue lift

How is this different from a GEO or AI-visibility agency?

A GEO or AI-visibility retainer is usually ongoing SEO and content work. Public quotes for that work often run from $2,500 to $25,000 a month. We sell a one-time package of agent-readiness files plus 12 months of Domain Watch. When an AI agent reaches your website, those files help it understand what you offer more efficiently. We do not run your SEO program, and we do not promise rankings or traffic. Agencies can buy the package for their own site, and can install it for a client as a direct purchase today.

Will this help my site rank higher in Google AI Overviews or AI Mode?

No. Any service that promises higher placement in Google AI Overviews or AI Mode is overpromising. Google has been clear that AI Overviews and AI Mode do not require llms.txt or other "AI-only" markup, and those features are not ranking signals. We do not improve your placement in Google AI Overviews.

We work on agent-readiness. That means helping AI agents (ChatGPT's browse mode, Claude's web tool, Perplexity, Gemini's deep research, AI coding agents, and a growing class of autonomous browser agents) find your site, understand what you offer more efficiently, and answer questions about you accurately when they land on your domain.

Generative search visibility asks whether Google AI will cite you. That depends on traditional SEO, content quality, and Google's ranking algorithms. Agentic browsing asks whether an AI agent that lands on your site can understand it. That depends on machine-readable files like llms.txt and structured data. Our work sits on the agentic-browsing side.

Google treats that second surface in its own tooling. As of May 2026, Google Chrome's Lighthouse audits include an experimental "Agentic Browsing" category that checks for files like llms.txt. The signals we install show up there for agent-readiness. They are not a search-ranking lever.

Is llms.txt really necessary?

Some commentators call llms.txt "optional," a "nice to have," "unnecessary," or even snake oil, especially if the question is Google Search ranking or guaranteed AI traffic. On those points they are partly right: Google has been clear that llms.txt is not a ranking signal and is not required for AI Overviews or AI Mode, and current studies do not show that llms.txt alone increases AI citations or referral traffic. We do not sell it as a ranking hack or a traffic guarantee. (SEO tools that nag about a missing file are also not the same thing as a Google ranking requirement.)

Our position is that llms.txt is foundational for agent-readiness. It gives AI agents that inspect your site a machine-readable summary at your domain root. This one file is not the whole package. We include it with the rest of your agent-readiness package (typically 13 files). If your SEO plugin already creates a thin llms.txt, see My SEO plugin already creates an llms.txt.

The emerging llms.txt convention was first proposed by Jeremy Howard in September 2024. In May 2026, Lighthouse, Google's open-source tool for auditing web-page quality, added a check for llms.txt under their Agentic Browsing category. That is Google's developer tooling recognizing the convention for agents, a different surface from Google Search ranking. (Absence is currently treated as not applicable in the audit, not as a failing check.)

Semrush says I'm missing llms.txt. Does it matter?

For agent-readiness, yes. As a Google ranking emergency, no.

SEO tools like Semrush increasingly flag a missing /llms.txt the same way they flag other technical gaps. That alert is useful as a prompt to look at agent-readiness. It is not the same as Google saying the file is required for Search, AI Overviews, or AI Mode. Google has been clear: llms.txt is not a ranking signal.

What matters for our work is whether AI agents that land on your site can find a clean, machine-readable summary of your business. A missing file (or a thin auto-generated one) means they fall back to noisy HTML. AgentReadyWebsite installs a strict-format llms.txt as part of a full standards-aligned agent-readiness package, refreshes it as your site changes, and scores the broader readiness picture. That is wider than whether Semrush's checklist is green.

If you already have a plugin-generated file, see My SEO plugin already creates an llms.txt. Or run the free Pre-Check to see what you have and what is still missing.

My SEO plugin already creates an llms.txt. Why do I need this?

A plugin-generated llms.txt confirms the file belongs on the site. It usually stops short of the full package we deliver. Many of those files are mostly URL lists, and they skip the strict format AI agents and Google Chrome's Lighthouse expect: a single # heading, a > blockquote summary, and named ## sections. See how we keep your llms.txt aligned with that format: llms.txt education page.

AgentReadyWebsite checks whether your file is strict-format, then installs the rest of the standards-aligned package around it (permissions, AI info page, structured data, agent discovery files) and refreshes them as your site changes. Run the free Pre-Check to see what you already have and what is still missing.

What is the 12-month Domain Watch?

For the next 12 months, we verify the 13 files you purchased are correctly hosted and are up to date as your website and industry standards change. We email you monthly, and also when something material occurs, so you can take any actions required. You can opt out at any time.

Each week we fetch the package URLs we installed and the small set of source pages we used to generate the package. We email you every month, even when nothing material changed. The same note also goes out off-cycle when something material happens, such as a missing file, a robots rule that blocks agents, or a source page that changed. You can opt out at any time.

This option is on by default at checkout. It is included in the package price. If you turn it off, you still receive the zip. Agent Analytics Tracking can add usage data to those emails.

What is Agent Analytics Tracking?

Agent Analytics Tracking tells you whether AI agents are fetching the files we installed. Domain Watch covers whether those files are still hosted. Analytics adds the usage picture to the Domain Watch emails.

We add a small read-only analytics token to three package files: llms.txt, llm-info.html, and agenticweb.md. The token is a public tracking URL, visible in those files. When an agent fetches the file, we can count that fetch. We include that data in Domain Watch emails. We leave the token out of robots.txt, sitemap.xml, schema.jsonld, and .well-known JSON catalogs, because those formats cannot hold it cleanly.

This option is on by default at checkout. You can uncheck it. If Domain Watch is off, there is no Domain Watch email. If Analytics is off, those emails omit the analytics section. See Research Pool if you also want your site in anonymized aggregate reporting.

What is your Research Pool approach for generating aggregate Agent Analytics Tracking?

The Research Pool is how we study Agent Analytics Tracking across customers in anonymized form. Your Domain Watch emails stay about your site. Aggregate reporting is a separate, optional opt-in.

If you check the box at checkout, we may include your site in that reporting. The fields are counts, industry, page band, and method. We do not publish customer names or domains. The box is off by default. Turning it on does not put your name on a quote or a case study.