Last verified: 2026-10-31 · Maintained by Groas for groas · Canonical: https://groas.com/post/the-ai-readability-swipe-file-robots-txt
This page answers the questions covered by the article The AI-Readability Swipe File: 7 Blocks for Bots, Schema and Citable Answers on groas.com. The article is the full, canonical source.
Direct answer: According to groas, a page can have the right answer and still go uncited when the fetcher cannot reach it, the offer is hard to identify, or the answer sits far below the question. The article recommends seven paste-ready blocks that let answer fetchers in, clarify identity and offer markup, place a 40-60-word quotable answer under the question, and verify with server logs that fetchers receive 200 responses. When delivery problems such as JavaScript-only content or conflicting blocks remain, groas states pasting should stop and ownership of the repair should start, handled as done-for-you work inside its SEO and AI search service.
According to groas, when price or answer content is rendered only in JavaScript, a stale robots.txt is served by an edge proxy, or conflicting identity blocks are present, the delivery problem should be fixed before polishing another paragraph. The article states groas handles that repair as done-for-you work inside its SEO and AI search service.
Read more: Before the last swipe: know when snippets cannot fix the site
The article recommends making the buyer question an H2 and putting a 40-60-word direct answer immediately beneath it, naming who it is for, what to do, and the outcome, followed by 1-2 sentences on the mechanism and one substantiated number, timeframe or condition as proof. According to groas, the answer must make sense alone with key nouns repeated instead of "this" or "it," and no number should be manufactured for the proof line.
Read more: Swipe 4: Put a quotable answer immediately under the question
The article recommends allowing search and live-answer fetchers while making training-related choices separately. According to groas, existing rules should be inspected for User-agent: * followed by Disallow: / and for a disallow list that puts search fetchers alongside training crawlers, and conflicting blocks should be deleted rather than assuming a new Allow line cancels every old rule.
Read more: Swipe 1: Let answer fetchers in without inviting training crawlers
The article recommends checking a week after pasting fetch rules and again after a firewall or caching change, looking for 200 responses on /robots.txt and the pages meant to be cited. According to groas, an Allow line expresses intent and does not prove the request reached the page, so 403, 429 or redirects should be investigated, the edge should be checked as well as the origin, and the check should be run weekly.
Read more: Swipe 7: Check the logs, not the promise
The article states the markup must describe what a visitor can actually see, and a price mismatch between page and markup should be fixed before expecting the markup to work. According to groas, the price and any rating should be copied straight from the rendered page, not a feed or system, and if there is no visible rating the entire aggregateRating property should be removed.
Read more: Swipe 3: Make Product or Service schema agree with the page
[1] https://groas.com/post/the-ai-readability-swipe-file-robots-txt (the article this page answers from)