How your AI can work with Hyperize.
Want your AI to assess our offering, find evidence or prepare a website check? hyperize.ai provides the same public content as HTML, Markdown and selected JSON versions. Four tools support research and enquiries. Access is free, with no API key or registration.
Check five technical features of your website
This quick check fetches a public website and looks for reachability, robots.txt rules, llms.txt, JSON-LD and Markdown. It does not perform a customer task or measure an Agent Success Score. Files or rules alone do not guarantee access or recommendations.
For agreed questions and actual task tests: Request a free Snapshot.
One site. Three formats.
Public claims stay the same. Protected reports do not become public through a different format.
We build integrations for customer tasks. This link explains the offering and its limits; it is not a list of completed platform integrations.
llms.txt guides you through the content. The German index links complete translations; /llms-full.txt holds the full public inventory.
For public content pages: append .md to the address or send Accept: text/markdown. The file header contains the title, description and original address.
Structured data for content listed in the site guide, including homepages, services, methodology, insights and the DAX 40 Index. Format: hyperize/v1. Not every page has a JSON version.
XML inventory of indexable pages. lastmod uses the content modification date, not the build date as a blanket substitute.
robots.txt describes access for search and AI crawlers. Permission proves neither retrieval, indexing nor citation.
JSON-LD in the HTML describes the organization, terms and, by page type, services, datasets or articles. The same published facts, classified for machines.
Unknown addresses return HTTP 404 with recovery links. Accept: text/markdown returns the error message as Markdown.
Three tools read. One requests.
A supporting browser exposes the tools through document.modelContext. Which tools your agent can use depends on its access. The linked content remains readable without WebMCP.
Retrieve a DAX 40 brand's measurement status and published values: score, AI Visibility, AI Usability and report link. Missing measurements remain identified as such.
Example: “How was adidas measured, and where are the barriers?”
Search complete studies and field tests in the current page language. An empty query returns the latest articles.
Example: “Find studies about AI agents shopping.”
Check five technical features of a publicly reachable website: front door, robots.txt, llms.txt, JSON-LD and Markdown. A surface probe, not a measured Agent Success Score.
Example: “Check example.com for these five features.”
Request the free Snapshot. Hyperize tests real agent tasks and emails the report with score and DAX 40 benchmark within 48 hours. Free, no commitment.
Example: “Request the Snapshot for my confirmed website and email address.”
request_snapshot sends a real enquiry through the same form a person uses. Have the user confirm the website and work email first. It does not place a paid order.
Six more tools on the homepages
English homepage · German homepage. An agent can adapt text to your context and create a decision brief. It reads allowed fields and length limits; accepted text appears immediately. Changes stay in the current tab. Reset or reload restores the original. The usage counter receives tool names, not visitor context or rewritten text.
get_adaptation_capabilities · start_personalization · write_all · write_section · show_decision_brief · reset_experience
Retrieve content directly.
# Site guide
curl https://www.hyperize.ai/llms.txt
# Services as Markdown: two ways
curl https://www.hyperize.ai/en/pricing.md
curl -H "Accept: text/markdown" https://www.hyperize.ai/en/pricing
# DAX 40 Index and an English brand report as JSON
curl https://www.hyperize.ai/en/dax40-index.json
curl https://www.hyperize.ai/en/dax40-index/brands/adidas.json
# Technical surface probe, not a measured score
curl "https://www.hyperize.ai/.netlify/functions/agent-readiness-probe?url=example.com"These examples read public content. The documented endpoints need no key. To discuss an integration with your systems: Discuss my integration.
A form can be reachable and still guide people poorly.
On 19 October 2026, we entered btdjb as the website in our Snapshot form and left the email empty. The form asked for an email instead of identifying the invalid website. No request was sent.
The revised form checks the website first, names the invalid field, associates the error with it and moves keyboard focus there. Correcting that field clears its error.
This example tests input validation on our own form. It does not measure agent recommendations, accepted requests, email delivery or customer results.
Try the form without submitting a request
