AI readability — what it is, and how to check yours.

Why now

Readable is the new crawlable.

Crawlable was the old bar: a bot could reach the URL. Readable is the new one: a machine can extract the meaning without a browser. The bar moved because the readers changed — the crawlers feeding AI assistants read the raw response and nothing else. A page can be perfect on screen and near-empty to them.

Most sites are not there yet. We measured 30 well-known websites: 56% publish an index for AI readers, and 7% serve anything readable behind it — see the benchmark.

Be readable, not just crawlable.

The checklist

Six checks, each one verifiable by a stranger.

• 1. The content is in the served HTML.What the server returns — before any script runs — contains the words that matter. Client-rendered content fails this for every major AI crawler.Verify: curl the page and search the response for a sentence you can see in the browser. Diagnose and fix invisible content.
• 2. A machine copy of each page exists.A clean Markdown version at a direct URL, so a reader gets the content without parsing your layout.Verify: Request a page with Accept: text/markdown — does anything answer? The Markdown-for-agents reference.
• 3. One index lists what matters.An llms.txt file: your key pages, each with a one-line description, curated for machine readers.Verify: curl https://yoursite.com/llms.txt What is llms.txt.
• 4. The facts are structured.Valid schema.org JSON-LD stating what the page is — product, article, business — making it eligible for rich results in Google.Verify: Run the page through Google’s Rich Results Test, or the Seogence analyzer. SEO & schema, explained simply.
• 5. AI crawlers are allowed — deliberately.robots.txt names the AI agents and reflects a decision per job: training, search index, live user fetch.Verify: Read your robots.txt against the current crawler roster. The AI crawler roster.
• 6. The machine copy is discoverable.A <link rel="alternate" type="text/markdown"> in the head points readers at the index, so the copy is found without guessing.Verify: View source on any page and look for the alternate link.

# If the words are not in the response, a no-browser reader cannot see them
curl -s https://yoursite.com/product | grep "a sentence from the page"

# Does your site have an index for AI readers?
curl -s https://yoursite.com/llms.txt

Do it yourself

Every check is passable by hand.

Server-side render the content. Build a Markdown export for every page. Write and maintain the llms.txt index. Author valid JSON-LD per page type and keep it valid as pages change. Audit robots.txt against a roster that keeps growing. Add the discovery tag to every template. Each one is a known job — together they are a build, and three of them are never finished, because content changes.

With Seogence

One tag serves the readable layer. The analyzer checks the rest.

Add the Seogence tag once and the Knowledge Layer serves checks 2, 3 and 6: a clean Markdown copy of each page at a direct URL over plain HTTP, a per-domain llms.txt index that re-reads itself when your content changes, and the discovery link that points readers at it — on WordPress, Shopify, Wix, Next.js or plain HTML, with no rebuild. Check 4 ships alongside: valid schema.org JSON-LD, eligible for rich results in Google. And you do not have to rebuild for check 1 — the readable copy is served beside your page, in a form that needs no browser.

Checks 1 and 5 stay yours — your stack, your policy. The analyzer shows you where they stand.

The honest line

Readability is what you control.

The checklist covers what a machine can read — that part is yours, and every item on it is verifiable by a stranger. What any assistant does with what it reads is theirs, and nobody can promise it. Serving your site in a readable form is the part of the outcome you own.

Questions

Common questions.