How ready is your site for AI agents?
ChatGPT, Claude and Perplexity read websites with their own bots when they answer a question. Enter your address and we will show you, with evidence, whether those bots can get into your site, read its content and identify you correctly.
What do we measure?
34 automated checks in 6 categories, plus a 23-item manual checklist for things that cannot be measured from outside (not part of the score). AI search bots: OAI-SearchBot and ChatGPT-User (ChatGPT), Claude-SearchBot and Claude-User (Claude), PerplexityBot and Perplexity-User (Perplexity).
6 checks
Agent access
Can AI search bots get into the site and index it?
- robots.txt is readable
- AI search bots are allowed
- Model training bots (your choice)
- Bots can actually get in (WAF)
- Home page can be indexed
- Sitemap is up to date
10 checks
Machine readability
Can the page be read without running JavaScript, and is its structure clear?
- Content is readable without JavaScript
- Page title and description
- Heading hierarchy
- Page language is declared
- Alt text on images
- Canonical URL
- Semantic HTML
- Page titles are unique
- Content date is shown
- Share preview (Open Graph)
8 checks
Structured data
Is it stated clearly with schema.org who you are and what you offer?
- JSON-LD is valid
- Organization identity (Organization)
- WebSite schema
- Service description (Service)
- Product and price (Product + Offer)
- Frequently asked questions (FAQPage)
- Contact details in the schema
- Breadcrumb trail (BreadcrumbList)
2 checks
AI files
Are there summary files for models, such as llms.txt?
- llms.txt
- llms-full.txt
2 checks
Agent protocols
API/MCP endpoints agents can act through (depends on site type)
- MCP server listing
- API description (OpenAPI)
6 checks
Technical basics
HTTPS, response time, error rate, compression
- HTTPS and redirect
- Server response time
- Inner pages load without errors
- Contact and privacy pages load
- No insecure resources on secure pages
- Compression (gzip/brotli)
How do we score?
Score
Each check has a weight of 1, 2 or 3. A passed check gets full points, a partial check half points and a missing check zero. Checks that could not be measured and checks that are out of scope for your site type are left out of the calculation. Every result comes with evidence of what we found.
Site type
- Service / corporate: Agency, clinic, law firm, consultancy, manufacturer or B2B company site
- E-commerce: A store that sells products; product schema and price/stock data are also checked
- Software / SaaS: API and agent protocols (MCP, OpenAPI) are also checked
Gate rules
If something critical is missing, the score is capped no matter how good the other checks are:
- At most 25: The home page is excluded with noindex → answer engines cannot use it as a source.
- At most 40: AI search bots cannot get in because of robots.txt or a firewall.
- At most 50: Content is not visible without JavaScript → most bots read an empty page.
Grades
- 90–100 · Exemplary: All core signals we measure are in place.
- 80–89 · Strong: Agents read the site easily; a few improvements remain.
- 60–79 · Ready: Basic access is in place; structured data and content signals need strengthening.
- 40–59 · Basic: Bots get in but struggle to understand the site.
- 0–39 · Invisible: AI engines cannot read the site or use it as a source.
What this analysis does not measure
- Visibility in AI answers. Whether an engine recommends you is a separate question from technical readiness, and a single measurement cannot prove it.
- Speed in the browser (LCP). Response time is the HTML download time measured from our server; it is not a real user experience measurement.
- The whole site. We check the home page and the first three inner pages in the sitemap; we do not crawl every page.
- Internal processes. Things that are not visible from outside, such as contracts, stock, payments or customer service, are not measured.
- How a firewall treats the real bot. If our request with a bot identity is blocked, we say so; some firewalls verify the real bot by its IP address and let it in.
Reading the report is easy; fixing it takes work.
We fix robots.txt and firewall settings, server-side content, schema.org, llms.txt and content structure together. Then we measure your visibility in AI answers on a regular basis.
Frequently asked questions
What does the AI readiness check measure?
It checks your site the way AI search and answer engines see it, in 6 categories and 34 checks: whether bots get past robots.txt and the firewall, whether the content can be read without running JavaScript, schema.org structured data, the llms.txt file and basic technical signals. For every check we write down what we found (the evidence).
How is the score calculated?
Each check has a weight from 1 to 3. A passed check gets full points, a partial check half points and a missing check zero. Checks that could not be measured or do not apply to your site type are left out; they neither lower nor inflate the score. Then gate rules apply: for example, if AI search bots cannot get into the site, the score is at most 40 regardless of the other checks.
What are the “Full checklist” and the “Fix plan”?
Next to the 34 automated checks, the report includes a 23-item manual checklist for things that cannot be measured from outside (Search Console and Bing verification, Google Business Profile, consistent company details everywhere, analytics setup and so on). Each item explains how to check it and how to fix it. Your ticks are stored only in your browser, for that report only, and do not change the score. The fix plan lists the gaps in order of importance, with the access needed, in one list; you can also copy each item as an instruction for a coding agent.
Why do I see the same result when I scan the same site again?
For the same site and site type, the latest report from the past 24 hours is shown. That way the same input gives the same result and the scanned site is not loaded unnecessarily. The measurement date is shown at the top of the report.
Does a high score mean ChatGPT will recommend me?
No. This analysis measures the technical groundwork AI engines need to read your site and understand it correctly. Being recommended in an answer also depends on your content, your position in search results and how your brand is mentioned in other sources. We measure that separately in our AI search optimization service.
Is llms.txt required?
No. llms.txt is a proposed standard (llmstxt.org) that introduces your site to models with a short Markdown summary, and it is not proven that all engines use it. Because it is easy to prepare, we measure it as a low-weight check.
Will the analysis put load on my site?
No. One analysis sends at most 18 requests in total, no more than 2 at the same time, and identifies itself as PikupHazirlikBot. We check your robots.txt rules before visiting inner pages. For the bot access test, one request is sent to the home page with each known AI bot identity.
What information do you store?
We store the scanned domain and the analysis result; these records are kept to improve our service and so you can reopen the report with its link. We do not ask for personal data. Your IP address is kept only as a one-way hash for abuse limits. There is no public list of scanned sites; only people who have the link can open a report.