Picture this: an AI assistant gets asked, "Who's the best HVAC company in Scottsdale?" It goes hunting. It finds your client's site. It reads the homepage, maybe a service page. And then it leaves — not because the site was bad, but because there was nothing to confirm what the site claimed.
No corroboration. No proof of existence. Just assertions.
That's the gap most agencies aren't filling yet. And it's costing their clients recommendations they should be winning.
Your SEO instincts are right about one thing: don't go tearing up client pages every time a new optimization trend surfaces. Wholesale content changes to rank for AI can absolutely damage the Google rankings you've already earned. That's not a trade worth making.
But AI assistants aren't crawling sites the way search spiders do. They're not just reading pages — they're trying to verify entities. They want to know: does this business actually exist, is it stable, and can I find proof of that beyond what the business itself says?
Among the AI ranking factors we test, business legitimacy and third-party corroboration consistently carry serious weight. An AI that can't verify a business is a real, stable, established entity is not going to stake its recommendation on it. That's the whole game for local right now.
The llms.txt standard gives you a clean, non-destructive place to lay out exactly what AI should know about a business — without touching a single ranked page.
Think of it as a dedicated briefing document for AI crawlers. It sits at the root of the site, it's publicly accessible, and it's specifically designed to be read by large language models. You're not gaming the algorithm. You're just making it easier for AI to do its job — and making your client the obvious, confident answer.
Here's the key insight: most of what AI needs to verify a business isn't on the website at all. It's off the site. Your job is to build the bridge.
When an AI crawler hits a well-built llms.txt file and follows even two or three of these signals, it starts building a picture of a real, verified, rooted entity. That confidence is what drives recommendations.
Here's what belongs in that file for your clients:
None of this is sensitive information. But a word of caution: anything you put in llms.txt is publicly readable, so counsel clients accordingly. Keep it factual, verified, and linkable — not anything proprietary or internal.
Here's how J. Brent thinks about it: every time an AI can read something on a client's site and then confirm it somewhere else, the confidence score for that entity goes up. The business license exists in the state database. The award exists on the newspaper's site. The address exists in the professional association directory.
That pattern — claim, then confirmation — is what makes AI think, okay, this is a real entity I can stake a recommendation on.
For local businesses especially, this is where the ranking battle is being fought right now. When someone asks an AI assistant for the best florist or plumber or Italian restaurant in their neighborhood, the businesses with the deepest corroboration web are the ones getting named. Not necessarily the ones with the prettiest websites.
Building a solid llms.txt file for every client manually takes time. That's why we built a free llms.txt maker — it pulls what already exists on the client's site, structures it in the order AI crawlers actually care about, and generates a ready-to-deploy file.
But here's the part that matters most to agencies: it doesn't just generate the file. It tracks when AI crawlers actually visit it. You get reporting that tells you when an AI system crossed through that page — real data, not guesswork, that you can put in front of a client.
That's a conversation-changer in a client review. "Here's when Google's AI crawler visited your llms.txt file last week." Try getting that from anyone else.
The tool is free for any business. If you're running AEO or GEO services and want to use it at scale for your clients, it's built for exactly that workflow. Find it at llmsmaker.com.
Don't rewrite client pages chasing AI. Build the corroboration layer that AI is already looking for. A well-constructed llms.txt file, loaded with verified, linkable proof points, is the lowest-risk and highest-signal move available to your clients right now.
Make their site sticky for AI. Give it somewhere to land that actually confirms what they're claiming.
That's the honey.