Free AI-readiness scantalabat.com
talabat.com
Hard for AI to use
AI readiness
Assistants can reach you, but they have to guess too much — which is when they name a competitor instead.
The headline is the sum of the six pillars below. These three tiers are the check breakdown: they score the same site on a different scale, so the two totals do not have to match.
Critical access
The four ways a site blocks an assistant before content even matters.
Assistants can reach the site
All 7 primary AI crawlers are allowed by robots.txt.
2 / 2 passed
Content exists without JavaScript
Raw HTML has 297 characters of readable text; H1 present, only 297 words.
2 / 2 passed
Missing pages fail safely
Either no sitemap.xml at /sitemap.xml or it returned 404.
1 / 2 passed
Controls are understandable
0/1 form controls have a label (0%).
3 / 4 passed
The six pillars
Headline 38 / 100 = crawler access 25 + llms.txt 0 + entity 0 + answers 3 + brand 5 + freshness 6.
- Crawler access25/25
- llms.txt0/10
- Entity & structured data0/20
- Answer readiness3/18
- Brand clarity5/15
- Freshness & trust6/12
Fix these first
Failures first, then partials, biggest point loss at the top. Each card copies the prompt that asks a coding agent for exactly that fix.
- 01
Form control labeling
InteractionWhat this checksWhether the controls an agent meets expose real names and roles instead of anonymous clickable boxes.−4Associate a visible <label for> with each input, or add aria-label.
- 02
Sitemap with lastmod dates
NavigationWhat this checksHow the site behaves when an agent takes a wrong turn: real 404s, honest redirects, a sitemap with dates, and markdown negotiation.−4Publish an XML sitemap at /sitemap.xml listing every indexable URL with <lastmod> dates.
- 03
JSON-LD entity identity
Entity and schemaWhat this checksThe JSON-LD that tells an assistant who you are: Organization, WebSite, sameAs profiles, and the content types on the page.
Every check behind the score
24 applicable checks. Each one shows the evidence the scanner read and, when it did not pass, how to pass it.
Essential
- AI crawler reachabilityPassed12 / 12
All 7 primary AI crawlers are allowed by robots.txt.
- Content without JavaScriptPassed8 / 10
Raw HTML has 297 characters of readable text; H1 present, only 297 words.
- Agent-friendly 404sPassed12 / 12
A nonexistent path returned a real HTTP 404 with a recovery body (sitemap/llms.txt links).
- Markdown content negotiationPartial 20%2 / 10
Accept: text/markdown returned text/html; charset=utf-8.
How to pass. Serve a markdown variant (Content-Type: text/markdown; charset=utf-8) when the client asks for it, and add Accept to Vary.
- Accessible document structure
Agent journey
One agent, eight fetches at most: it reads the site and explains it back.
Agent is reading the site…
Does ChatGPT actually recommend talabat.com? PRO tracks it weekly.
The same questions your buyers ask, put to ChatGPT, Perplexity, Gemini and Claude every week, with who they name instead. January 2027, $10 a month — not on sale yet.