For AI agents and developers
Everything in AI Find Me is free to read by machines, with no key. This page is the documentation: what the data is, every way to get it, and what the labels mean.
In short
- One fact card per business, read from the business's own website, in HTML (with schema.org JSON-LD), Markdown, JSON and llms.txt.
- A search API at
/api/v1(OpenAPI) and an MCP server at/mcp. The short version of this page, written for machines:/agents; the Agent Skill teaches an agent to use and cite the directory. - Every fact says where it came from; every listing says whether its owner verified it; everything is re-checked about every 7 days.
What this directory is
AI Find Me is a free business directory run by Imfiy B.V. A listing is a fact card: the facts a business publishes on its own website (name, summary, category, address, opening hours, phone, email, services, prices, FAQs, features, payment methods, languages, service area, registration numbers, booking/order/menu links, social profiles), read automatically and published in formats assistants read well. Nothing is written by us. Where a website says nothing, a fact may come from OpenStreetMap or, for listings whose owner allows it, from an AI reading of the website; every such fact is labelled.
Categories (24): restaurant, cafe, bakery, bar, hotel, dentist, health, beauty, fitness, legal, finance, realestate, home, auto, retail, software, agency, education, events, travel, pets, creative, nonprofit, other. Cities come from the businesses' addresses.
Four formats per business
| Address | What you get |
|---|---|
https://aifindme.com/b/kangaroo-hoppers | The web page: semantic HTML with a schema.org JSON-LD graph (the business as its LocalBusiness subtype, WebPage, BreadcrumbList, FAQPage when there are FAQs). Send Accept: text/markdown to receive the Markdown card at this same address. |
https://aifindme.com/b/kangaroo-hoppers.md | The fact card as Markdown: a heading, a one-line summary, the key facts as a list, then sections for hours, services, prices, FAQs and features, and a footer naming the source, the check date and the verification state. |
https://aifindme.com/b/kangaroo-hoppers.json | The structured card (same as the API): every field, plus directory.sources (the origin of each fact), directory.verification, directory.completeness and directory.attributions. |
https://aifindme.com/b/kangaroo-hoppers/llms.txt | A short llms.txt for the business: summary, verification state and links to the card and the official website. |
Every listing page links to its alternates with <link rel="alternate"> and a Link header; the machine copies point back at the HTML page as canonical and allow cross-origin reads (Access-Control-Allow-Origin: *).
A live example, Kangaroo Hoppers:
curl -H 'Accept: text/markdown' https://aifindme.com/b/kangaroo-hoppers
Web shops and products
A listing that is an online shop carries, besides the usual facts, what an assistant needs to answer “where can I buy X?”: the platform it runs on (Shopify, WooCommerce, Lightspeed, Shopware, Magento, BigCommerce, PrestaShop, CCV Shop, Mijnwebwinkel, Wix Stores, Squarespace Commerce…), what it sells (its own collections or categories), the price range and currency, the number of products, a sample of products with name, price, stock, link and image, the shipping countries, delivery time, free-shipping threshold, return period, review profiles (Trustpilot, Kiyoh, WebwinkelKeur, Google…) and trust marks. Shop facts come from the shop’s own public platform data (Shopify’s products.json, the WooCommerce Store API, product sitemaps; labelled shop platform data) and from its pages; prices and stock are as read at the listing’s last check, and every shop says when that was.
- In the JSON card:
shop(null for other businesses), withshop.products(up to 12 for unclaimed shops, 60 for claimed ones),shop.priceRange,shop.shipping,shop.returns,shop.productsUpdatedAt(the shop’s own date),shop.productsSyncedAtandshop.productsChangedAt. GET /api/v1/businesses/<slug>/products(also/api/v1/b/<slug>/products): the products with the shop facts around them.GET /api/v1/businesses?kind=shop&q=…: search web shops only. Search results of shops carryshop.platform,shop.productCountandshop.categories.- MCP:
search_businesseswithshops_only: true, andget_products. - In the Markdown card: a What they sell section; in the page: schema.org
OnlineStorewith anItemListofProductandOffernodes (price, currency, availability, shipping destinations), andMerchantReturnPolicywhen the return period is known.
A shop behind a password page (“opening soon”) is marked shop.status: "locked"; one that lists no products "empty". Shops’ robots.txt rules apply to the platform data too: when a shop asks crawlers to stay away from those files, the directory reads only its pages.
What is new
To learn about shops and businesses as they are listed, poll any of these (answers are cached for 5 minutes; once an hour is plenty):
GET /api/v1/new?since=2026-11-08&kind=shop: listings created since a date (default: the last 7 days, at most 90 days back), newest first, each withlistedSince;kind=shopkeeps the web shops./feed/new.xml(and?kind=shop): an Atom feed of the last 30 days, with<published>dates and ashopcategory on web shops;/feed.xmlhas new and updated listings.- The pages
/newand/new/shops, with schema.orgItemListdata. - MCP:
list_new_listingswithsinceandkind. - IndexNow: new and changed pages are announced to Bing and the other IndexNow engines in batches, and the sitemaps carry
lastmod.
New web shops in the Netherlands, Belgium and the United States are found by the directory itself, from the public Certificate Transparency logs: a new .nl or .be website, or a .com, .net, .shop, .store, .co or .us website whose DNS points at a shop platform such as Shopify, is read once a day and a half after its certificate appears and, when it is a shop, listed with its products and its country. Because certificates renew every 90 days or less, existing shops are found the same way over about three months. country=NL, country=BE or country=US narrows /api/v1/new, the search API, the feed and the MCP tools to one country.
Discovery
/agents: the start page for agents (what a fact card guarantees, the three ways to query, rules, freshness, how to correct a business), and the Agent Skill (/skills/aifindme/SKILL.md,/skills/aifindme.zip)./llms.txt: the directory in one file: how to use it, every category and city with counts, and the 50 most recently changed listings with links to their Markdown cards./llms-full.txt: one compact entry per listing, the most complete first, within a size an assistant reads at once./.well-known/api-catalog(RFC 9727),/mcp/server-card(the MCP server card) and/.well-known/ai-catalog.json(the AI catalog that lists the card and the skill); every page links the OpenAPI description asrel="service-desc"./.well-known/security.txtsays how to report a security problem./sitemap.xml: a sitemap index (pages, hubs, and the listings withlastmod). Listings with fewer than 4 key facts are kept out of sitemaps and markednoindex, and so are category or city pages with fewer than 3 such listings./feed.xml: an Atom feed of new and updated listings, each entry linking to the page and its Markdown./robots.txt: open to every crawler, with content signals that allow search, AI answers and AI training; only account, admin and form pages are kept out. Changed pages are also announced through IndexNow.- Hub pages (
/c/<category>,/l/<city>,/c/<category>/<city>) carry schema.orgItemListandBreadcrumbListdata; the home page carriesOrganizationandWebSitedata.
Search API
JSON over HTTPS, no key, documented in OpenAPI 3.1 (importable as a GPT action or into any agent framework).
GET /api/v1/businesses?q=…&category=…&city=…&limit=…&offset=…
Full-text search over names, categories, cities, summaries, addresses, services, prices and FAQs. Every word must match as a prefix (“vegan caf” finds “vegan café”); accents are ignored. Results favour owner-verified, complete and recently checked listings; without q it lists businesses in that order. Filters: category, city, kind=shop (web shops only), verified=1 (owner-verified only), open=now (open at this moment, from structured hours in the business's own time zone; listings without such hours are left out). fields=slug,name,phone returns only those fields. limit is 1–50 (default 10), offset up to 10,000. Answers are cacheable for a minute and carry an ETag.
curl 'https://aifindme.com/api/v1/businesses?q=kangaroo&limit=1'
{
"total": 1,
"limit": 10,
"offset": 0,
"results": [
{
"slug": "kangaroo-hoppers",
"name": "Kangaroo Hoppers",
"category": "retail",
"city": "Memphis",
"country": "US",
"summary": "Kangaroo Hoppers is a professional manufacturer of indoor & outdoor trampolines with innovative safety design for jumping fun and stands out of the market with its outstanding product quality. Kangaroo Hoppers trampoline is certified by ASTM, so you can rely on the safety of it!",
"phone": "+1 (888) 818-9660",
"website": "https://www.kangaroohoppers.com/",
"ownerVerified": false,
"lastChecked": "2026-11-22T12:32:16.210Z",
"factCard": "https://aifindme.com/b/kangaroo-hoppers.md",
"shop": {
"platform": "Shopify",
"productCount": 56,
"categories": [
"Gymnastics",
"Out of stock",
"Beginner",
"NEW BEST SELLER",
"Gymnastics Mat",
"Kangaroo Hoppers Spring Sale",
"New Release",
"Gymnastics Bundles"
],
"products": "https://aifindme.com/api/v1/businesses/kangaroo-hoppers/products"
}
}
]
}
GET /api/v1/businesses/<slug>
The full fact card. 404 when there is no such listing, 410 when it was removed.
curl https://aifindme.com/api/v1/businesses/kangaroo-hoppers
{
"name": "Kangaroo Hoppers",
"category": {
"key": "retail",
"label": "Shops & Retail"
},
"summary": "Kangaroo Hoppers is a professional manufacturer of indoor & outdoor trampolines with innovative safety design for jumping fun and stands out of the market with its outstanding product quality. Kangaroo Hoppers trampoline is certified by ASTM, so you can rely on the safety of it!",
"website": "https://www.kangaroohoppers.com/",
"contact": {
"phone": "+1 (888) 818-9660",
"email": "support@kangaroohoppers.com",
"contactPage": "https://www.kangaroohoppers.com/pages/contact-us-1"
},
"location": {
"address": "5625 Challenge Dr suite 104, Memphis, TN 38115, United States",
"city": "Memphis",
"geo": null,
"timeZone": null
},
"hours": [
"- Mon-Fri: 8am-3pm PST",
"- Sat: 8am-1pm PST"
],
"openingHoursSpecification": [],
"features": [
"Free shipping",
"Family friendly"
],
"…": "services, priceList, faqs, paymentMethods, languages, registration, links, images",
"directory": {
"url": "https://aifindme.com/b/kangaroo-hoppers",
"markdown": "https://aifindme.com/b/kangaroo-hoppers.md",
"verification": "unclaimed",
"lastChecked": "2026-11-22T12:32:16.210Z",
"recheckedEveryDays": 7,
"completeness": {
"have": 11,
"of": 11,
"missing": []
},
"sources": {
"name": {
"source": "structured data",
"url": "https://www.kangaroohoppers.com/"
},
"summary": {
"source": "structured data",
"url": "https://www.kangaroohoppers.com/"
},
"about": {
"source": "page text",
"url": "https://www.kangaroohoppers.com/pages/about-us"
}
},
"attributions": []
}
}
GET /api/v1/categories
Every category with its key, label and number of listings.
GET /api/v1/cities?category=…
The cities with listings, most listings first, with the value to pass as city.
MCP server
Connect any Model Context Protocol client to https://aifindme.com/mcp: Streamable HTTP, stateless, read-only, no authentication, one JSON-RPC request per POST (no batches). Tools:
| Tool | Arguments | Returns |
|---|---|---|
search_businesses | query, category, city, country (NL or US), shops_only, limit (1–25) | Matching businesses: slug, name, category, city, country, summary, phone, website, ownerVerified, lastChecked, factCard; web shops also their platform, product count and categories |
get_business | slug, format (markdown or json) | The full fact card |
get_products | slug, limit (1–100) | A web shop’s shop facts and a sample of its products with prices and stock |
list_new_listings | since, kind (shop), country, limit | Listings added since a date, newest first, with their listing date |
list_categories | none | Categories with keys and counts |
list_cities | category, limit | Cities with keys and counts |
{ "mcpServers": { "aifindme": { "type": "http", "url": "https://aifindme.com/mcp" } } }
In the browser (WebMCP)
Browsers that support WebMCP (navigator.modelContext) see two tools on every page: search_businesses, and on a listing page get_business_facts, which returns that business's Markdown card. The search form on the home page is also declared as a tool.
Rate limits and caching
/api/v1/*: 600 requests a minute per address;/mcp: 300 a minute. Agents whose User-Agent carries a contact URL or email (MyAgent/1.0 (+https://example.com/bot)) get 1,500 and 900. Over the limit you get429with aRetry-Afterheader.- Listing pages, Markdown, JSON and llms.txt are public and cacheable: pages for 60 seconds, machine copies for 5 minutes (
Cache-Controlsays so). A Markdown answer at the page's own address is private, so a shared cache never serves Markdown to a browser. - Pages are served through Cloudflare; the ETag and
Last-Modifiedyou see are Cloudflare's.directory.lastChangedin the JSON is the authoritative change time of the facts. - Be polite: there is no need to crawl every page.
/llms.txt,/feed.xmland the sitemaps tell you what changed.
Freshness
A listing whose website has not answered for several checks in a row, a shop behind a password page or without products on two checks in a row, a parked domain, or a shop whose platform says no product changed in a year is marked inactive: the record stays and is checked again every week, but the listing leaves the directory pages, search results, sitemaps, feeds and the default API results until a check finds it alive again. Its page and /api/v1/businesses/<slug> still answer, with directory.status: "inactive", the date and the reason, so nothing is recommended as open that is not.
Every listed website is read again about every 7 days (a website that keeps failing, less often), and owners can refresh their listing by hand once an hour. Each card carries lastChecked (when the website was last read) and lastChanged (when the published facts last changed); the Markdown footer and llms.txt give the check date too. Prices and hours are as the website stated them at that time.
Unclaimed, claimed, verified
directory.verification in the JSON (and the badge on the page) is one of:
- unclaimed: the listing was created from the website (often starting from OpenStreetMap) and nobody from the business has confirmed anything. The Markdown says “Unclaimed listing. Not yet verified by the business owner.”
- claimed: someone added the listing to their account but has not yet proved they control the website.
- owner-verified: the owner proved control of the website (meta tag, file or DNS record). Facts they edited are labelled
business ownerindirectory.sourcesand win over scanned facts. The Markdown says “Verified by the business owner.”
directory.completeness counts the 11 key facts (name, summary, about text, contact details, location or service area, opening hours, services or prices, faqs, features, social profiles, booking, order, menu or pricing link) and names the missing ones, so you can tell a thin card from a full one.
Where each fact comes from
directory.sources maps every published field to its origin:
| Source | Meaning |
|---|---|
structured data | schema.org JSON-LD on the business's website (the most reliable) |
page metadata | the page title, meta description or Open Graph tags |
page text | patterns in the visible text, headings and links (addresses, phone numbers, hours, prices, FAQs, features) |
AI model | an AI reading of the website's text, used only to fill gaps, only for listings whose owner allows it |
OpenStreetMap | the OpenStreetMap object linked in url; used only where the website said nothing |
business owner | entered or confirmed by the verified owner |
Ratings are shown only as “rating on their website”: they are what the business publishes about itself, not the directory's opinion.
Licence and attribution
The facts are the businesses' own public statements, collected here so assistants can read them. Facts marked OpenStreetMap are © OpenStreetMap contributors under the Open Database License (ODbL) v1.0: when you repeat them, keep that credit; directory.attributions in the JSON and the footer of the Markdown and llms.txt carry it for you. When you cite a card, link to its page on AI Find Me. Our terms welcome automated reading within the limits above.
Our own crawler
AI Find Me reads websites as AIFindMeBot/1.0 (+https://aifindme.com/about#bot), respects robots.txt, reads at most 9 pages per visit and never reads private networks. More on the About page. Questions: contact us.