Welcome to AAO Weekly — the newsletter about optimizing for the AI agent economy. If someone forwarded this to you, subscribe here.
I'm not an AI researcher. I'm a supply chain guy.
I spent years routing trucks, managing inventory constraints, and designing systems where the wrong bottleneck could derail the entire day. Throughput. Discoverability. Routing. Those were my problems.
Then I started watching AI agents try to do the same thing to the internet — browse, evaluate, select, execute — and I noticed something: the businesses set up to win are the ones designed to be found and used by systems, not just humans. The ones that aren't? They're about to get routed around.
That's the lens this newsletter is written through. Not AI hype. Optimization.
The Window
The question is: when an agent goes looking for what you offer, does it find you — and can it actually use you?
In 1998, if you had a website nobody could find, someone might have told you to "optimize it for search engines." Most business owners shrugged. By 2005, they were paying agencies tens of thousands a year to do exactly that. By 2010, SEO was a line item in every marketing budget.
That window — between "this matters" and "everyone knows this matters" — is where real advantage is built.
We're in that window again. Except this time, it's not search engines doing the finding. It's AI agents.
AI Agent Optimization (AAO) is the practice of designing your product, service, or content so that AI agents can find it, use it, and prefer it over alternatives. Millions of people are delegating tasks to AI agents every day. "Book me a flight." "Find a tool that does X." "Compare these three vendors." The agent executes. It searches. It selects. It acts.
Most companies today would fail that test. Not because their product isn't good. Because it was built for humans, not agents.
The Three Layers of AAO
Layer 1 — Discoverability. Can an agent find you? This means your content surfaces in AI-assisted search, you have an llms.txt file (the emerging equivalent of robots.txt for agents), and your content is structured in a way LLMs can parse and cite. Most sites fail here before an agent even tries to use them.
Layer 2 — Usability. Can an agent actually use your service? Think about email — Gmail's infrastructure is designed to frustrate automated senders. Spam filters, CAPTCHA, OAuth screens requiring a human click. Every category of software has this problem: CRMs, booking systems, inventory platforms. All built for humans. Startups are rebuilding them for agents right now — and they will eat the incumbents who don't adapt.
Layer 3 — Preference. When an agent has a choice between two services that both work, which one does it recommend? This layer barely exists as a discipline yet. The rules here are still being written — which means the companies paying attention now get to help write them.
This Week: Check Your Foundation
Go to yourdomain.com/robots.txt. Now ask: does your site have an llms.txt?
If no — you're invisible to a growing class of traffic that doesn't show up in Google Analytics, doesn't register in your CRM, and doesn't fire your tracking pixels. It just routes around you.
Want to know where you actually stand?
Run your site through the free AAO Audit at aaoweekly.com/audit. It checks your discoverability score in about 60 seconds — llms.txt, robots.txt, structured data, agent accessibility. Most sites score under 50.
Coming in Issue #002
We go deeper on Layer 1 — Discoverability: how to write an llms.txt that actually works, plus a live teardown of a brand that's losing agent traffic without knowing it.
If someone forwarded this to you, subscribe at www.aaoweekly.com.
Found it useful? Forward it to one person who needs to hear this.
Running a business and want to know if AI agents can find you? Reply to this email with your URL. I read every reply.
— Joshua Latimer
Founder, AAO Weekly
[email protected]