Your Site. Your Rules — Implement AOM on Your Website Today

The web started as pages for humans. Today, agents read those same pages to summarize, recommend, and act — often without a contract that says what they may do with your work.

The problem: your content, but not your control

Every site — from blogs to product catalogs — is being read, parsed, and repackaged by agents and LLMs. Much of that happens without clear permission, attribution, or alignment with how you want your brand and data used.

The early web had the same chaos with crawlers until robots.txt gave sites a simple, standard way to set boundaries. We are at a similar moment for agents: not just where they may fetch, but how they may interpret, reuse, and integrate what they find.

The next step after robots.txt

robots.txt told crawlers which paths to skip. It fit the search era. Agent Object Model™ (AOM) is the next layer: machine-readable policy and structured surfaces so you can say what agents are allowed to understand, do, and respect — not only whether a URL was fetched.

Think of AOM as a shared vocabulary for human–agent–web interaction:

• Declare automation posture — who may automate what, and under which policy (e.g. forbidden, allowed (with guardrails), or open), aligned with how you run the business.
• Publish structured surfaces alongside HTML so intent does not depend on layout, A/B tests, or fragile selectors.
• Offer verifiable context so summaries and derivatives are less likely to misrepresent you.
• Integrate trusted agents deliberately instead of treating every script as interchangeable.

Normative definitions and schemas live on agentobjectmodel.org (MIT). aom.tools is where you validate surfaces, grab plugins and kits, and wire this into your stack.

Your site, your rules

Without a declared model, the default is implicit: scrape, guess, and ship. With AOM, you publish the terms of engagement — read-only, reuse with attribution, no training on this corpus, or whatever your policy allows.

That unlocks scenarios such as:

• Product sites exposing specs and policies so trusted agents can quote accurately.
• Publishers stating attribution and licensing expectations before reuse.
• APIs and brand sites sharing a consistent, validatable contract instead of one-off scraper recipes.

Agent manners

Just as robots.txt normalized crawler etiquette, AOM aims to normalize agent etiquette: structured, declarative, and something validators and CI can enforce — not hand-wavy “best effort.”

Start today

AOM is designed to be additive: you keep your human-facing site; you add policy JSON and page-level surfaces where it matters first.

On aom.tools

• Getting started
• Plugins for common stacks
• Downloads for surface owners
• Validator for your JSON

On this site

For a deeper take on why structured surfaces beat HTML-only automation, see How to design websites for AI agents (not just humans). If you build or operate agents (not just sites), Know Your Agent (KYA): Why AOM Agents Need an Identity covers registration, keys, and accountability on the agent side.

---

It’s your site, your rules — publish them where agents can read them.