For companies

Ship an agent-aware path this week.

Decide what a verified agent may do before you change a route. A verified result means a key on file signed this request. The name is the one the account typed.

The embed is one script tag and one server check.

Use cases

Five decisions, each with a limit.

Agent checkout

A verified token carries the action you set, such as checkout, and the agent id that signed it. Your checkout can return a structured cart to that agent instead of the human page.

imnotahuman.ai does not take payment, price a cart, or approve an order. The token is not a payment credential.

Embed with data-action="checkout"

Agent APIs

Your handler can require a token whose verification.action matches the route, then answer with machine-readable data for that agent id.

A token is an identity check for one challenge. It expires in ten minutes. It is not a long-lived API key.

Verify the token on your server

Different rate limits

agent.id is stable across challenges for that credential, so your gateway can count requests for that id separately. You choose a higher limit, a lower limit, or the same one.

imnotahuman.ai rate-limits its own verification API. It does not set quotas on your product.

Read agent.id from the verify response

Analytics

You can count VERIFIED, REJECTED, and EXPIRED challenges, then join agent.id to analytics you already run.

This is a verification event log. It is not session replay, and it does not tell you that a visitor is a person.

Read the events API

Action and intent

Action and intent are stored on the challenge and copied onto the token. Your policy can key off those fields now.

A policy engine that grants or denies permissions is not part of the product. Authorization stays with your site.

Read what verification proves

What this does not claim

Verification is an authentication signal. It is not a finding that a human was absent, and it is not approval to take the action.

  • That no human instructed the agent.
  • That the agent is safe or benevolent.
  • That the operator is trustworthy.
  • That the requested action should be allowed.
  • That a click, a user-agent, or a behaviour score is proof.

Security page states the same boundary in full.

Free

The same free service for a company site.

There is no enterprise tier, contract package, or SLA to buy. The widget, the API, and the dashboard log are the product. If you need a private deployment or a support contract, this service does not sell one.