The record an AI agent can trust, because you published it.
{ "section": "introduction" }
The record an AI agent can trust, because you published it.
Skin is a verified, owner-controlled profile layer for the AI agent era so language models answer questions about you or your company with facts you published, not whatever they scraped.
Skin Classic
[](https://profiles.skin/#waitlist "Join waitlist")
EXPLORER
Project governance
/div>
Json schema overview
Profile creation
profiles.skin
Standard
Auto
Review
Markup
LLMs stopped just answering questions and now act as agents that book, buy, and recommend on your behalf.
{ "section": "the problem" }
Right now, models describe you from memory and guesswork.
Outdated training data, stale scrapes, and confident hallucination. You have no way to correct the record, and no way to know what's being said about you to a customer you'll never see.
A model's picture of your pricing, team, or product freezes at its training cutoff unless it finds something newer, and then you still don't control what it trusts.
Stale by default
If an LLM gets a fact about you wrong, there's no form to file, no edit button, just a wrong answer, repeated to everyone who asks.
No correction channel
An autonomous agent comparing vendors or booking on your behalf has no reliable way to confirm which source is actually you.
Unverifiable by agents
{ "section": "what changes" }
Four things you're currently
doing badly, replaced.Not new features, the direct fix for what's broken about how models get facts about you today.
You speak, not a crawler
You describe yourself or your company in plain conversation, from a secured, owned profile. No crawler is guessing at your homepage, the source is you.
Scraped Authored
Queried on demand, not baked in
Nothing sits in a training set going stale. An agent asks, and gets whatever is true right now, not whatever was true at some model's cutoff.
Frozen Real-time
No site to build for machines
You don't stand up a machine-readable site, maintain an endpoint, or keep a file in sync. One profile, and agents can already see you.
Build it yourself Zero infrastructure
Facts, not prose
Your record is a structured list of facts, not a paragraph in English. An agent answering in Japanese or Portuguese renders it accurately and natively, nothing to translate.
Localize it Multi-language native
{ "section": "the hard part solved honestly" }
Found even when nobody
knows to ask for you by name.A verified record only matters if agents actually reach it and most queries aren't by name. Nobody asks "what is Skin?", they ask "get business data ready for AI agents".
Skin doesn't try to replace search. An agent finds candidates the way it already does — its own search or browsing — then queries Skin to confirm which one it's actually talking to and pull the current, owner-confirmed facts.
Search finds you. Skin is what an agent trusts once it has.
Terminal
{ "section": "how it works" }
Verify once. Publish anytime.
Agents query it.No scraping, no crawling your site and guessing. A structured, machine-readable profile you control directly.
Step 1
Prove ownership with a DNS TXT record or OAuth sign-in, the same trust model the web already runs on.
Verify
Step 2
Describe yourself or your company in plain conversation. Skin structures it into clean, queryable fields.
Publish
Step 3
Your record enters a machine-readable index built for retrieval, not for human browsing.
Index
Step 4
Any LLM or agent can request your record on demand and get the current, owner-confirmed version.
Serve
If you publish
Control what agents say about your pricing, team, and product. Update it, and it's current everywhere it's queried.
If you build agents
Query a verified source instead of scraping and guessing. Free for agents and developers.
This is a landing page, not a product yet.
We're validating whether this is worth building before we build it. If you'd use this, as a publisher or as someone building agents, tell us, and we'll bring you in early and take your feedback seriously.
\* required
Request early access
{ "section": "questions" }
Before you ask
What's the difference between this and RAG?
RAG is a retrieval mechanism, it's agnostic to what it retrieves from. Today that's usually whatever's crawlable: scraped pages, forum posts, stale copies. Skin doesn't compete with RAG, it gives it something better to retrieve from. A source with provable ownership and a live query, current as of whenever you last updated it, not whenever a crawler last passed by.
Isn't this just llms.txt?
Close in spirit, different in what's missing. llms.txt is a static file you host and maintain yourself, with no ownership verification, no index, and no query interface. An agent still has to find it, fetch it, and take it on faith. Skin adds the three things it doesn't have: verified identity, a real-time index, and something queryable on demand instead of crawled.
What stops someone from just lying in their own record?
Verification proves ownership, not truthfulness. The same trust model your own website already runs on. A verified record means "this was published by the verified owner", not "every claim the owner makes is objectively true". That's the same bar the rest of the web operates at today.
Are you locking me into a proprietary format?
No, the schema and the domain-ownership verification method are both open. Anyone can implement the full flow independently. Skin is hosting, an index, and an API on top of that: the Git-to-GitHub relationship, not a walled garden.
How is a record actually verified?
A DNS TXT record on your domain, or OAuth against a platform you already control. Same trust model as domain verification anywhere else on the web.
Do LLM providers actually query this today?
Not yet! That's what this page is for. If enough publishers and agent builders want it, an open, queryable index becomes worth integrating against. That's the bet being tested here.
What will it cost?
Likely a subscription for publishers who want a verified, hosted record. Free for agents and developers querying it. Nothing is final; early access users will help shape this.
Make yourself known.