Do llms.txt files actually help with AI search? In the first episode of MAXVolume, Brian Winum weighs the skeptical case, examines the Google tooling signals he finds significant, and explains why clear brand information and useful content matter more than a file alone.
Key Takeaways
Show notes
Do llms.txt files actually work, or are they another SEO distraction? In the debut episode of MAXVolume, Brian Winum explains why his answer depends on what you expect the file to do. His argument is that a file can be a useful entry point to a business’s information, but it cannot do the work of the website behind it.
Brian starts with the basics: an llms.txt file is intended to give compatible AI tools a concise overview of a site and links to important content. He compares it to a table of contents, then considers the skeptical case: it is a proposed convention, Google Search does not require it, and one published crawl experiment found very few requests for the file.
The episode then turns to the signals Brian watches in Google’s developer tools. He discusses Lighthouse’s agentic browsing checks and Google’s Agent Development Kit documentation, and draws a comparison with the way he saw Core Web Vitals move into everyday SEO work. That comparison is his interpretation of the pattern, rather than a prediction confirmed by Google.
There is a practical distinction between these examples. Google Search says llms.txt does not affect its visibility or rankings, while Lighthouse treats the file as optional. ADK documents an explicit way to give coding tools access to its documentation. These uses should be evaluated on their own terms; they do not establish that every AI search product reads or rewards a website’s llms.txt file.
Brian’s strongest recommendation is to focus on the context behind the file: accurate business information, useful service pages, current FAQs, and evidence that supports what you say. A clear index pointing to weak or outdated content still points to weak or outdated content. The episode closes with his case for treating llms.txt as one part of a broader effort to make a website useful to people and compatible AI tools.
Chapters
Host
Brian Winum is a partner at MAXPlaces and MAXBURST with more than 20 years in digital marketing. On MAXVolume, he examines AI search, SEO, and what changes in online discovery mean for businesses.
About Brian Winum →Full transcript
1,219 words · Full transcript from supplied captions · Filename spelling normalizedTranscript provided from the episode subtitles. Select a timestamp to play from that point.
Google’s mixed signals on llms.txt
In May of this year, Google published documentation that basically told everyone in my industry to stop worrying about llms.txt files. You don't need them. Relax. Less than a week later — the same week — Google's own Chrome team added a check for llms.txt to their official audit tool. So which is it? That little contradiction is the most honest thing Google has told us all year, and it's exactly where I want to start.
Welcome to MAXVolume with Brian Winum
Hey, welcome to MAXVolume. I'm Brian Winum.
I've spent a little over twenty years in digital marketing — I'm a partner at MAXBURST on the web design side and MAXPlaces on the strategy side — and most of my attention lately goes to one question: what happens to search, and to your business, when the thing doing the searching stops being a person and starts being an AI.
That's what this show is about. I turn up the signal and turn down the noise on AI search. I'll decode the updates that matter, call out the hype that doesn't, and give you my honest take even when it isn't the popular one.
What I won't do is read you a press release. There are enough people out there repeating whatever Google said in a blog post and calling it strategy.
So let's start with the question I've gotten more than any other this year: do llms.txt files actually work? And the real answer is more interesting than either side will tell you.
What is an llms.txt file?
Quick version of what we're even talking about. An llms.txt file is a small text file you put at the root of your website. It hands an AI a clean, clutter-free summary of your site — here's who we are, here's what we do, here are the pages that matter most. Think of it as giving someone the table of contents instead of making them read the entire book just to figure out what it's about.
The skeptical case
Now, I want to be fair to the skeptics here, because they've got real evidence. Their case is three parts.
One: it's not an official standard. It's a proposal — a rule nobody actually agreed to enforce.
Two: Google itself said you don't need it, and they didn't whisper it. One of their own people compared it to the old keywords meta tag, which has been useless for about fifteen years.
And three, the strongest one: people watched their server logs, and the AI bots that supposedly care about this file requested it in roughly one-tenth of one percent of visits. They're walking right past it.
So if you stop there — and a lot of smart people do stop there — the verdict writes itself. Skip it.
Watching Google’s tools as well as its statements
Here's why I don't stop there. There's an old piece of advice in my world: don't listen to what Google says — watch what Google does. I've been saying some version of that for twenty years, and every single time Google downplays something in public while quietly building it into their tooling, I watch a fresh wave of business owners get caught flat-footed when the official word finally shows up.
So let's watch what Google is actually doing. That same week they told everyone to ignore llms.txt, their Chrome team shipped an update to their audit tool with a brand-new category built around AI agents — and one of its checks grades your site on whether you have an llms.txt file, and whether you built it correctly.
Sit with that for a second. One team at Google says the file doesn't matter. Another team at Google built a tool that grades you on it.
And it's not just the audit — Google's own developer toolkit for building AI agents ships with llms.txt files baked right in. Google is optimizing Google's own properties for this. That is about as clear a tell as you are ever going to get.
Brian’s Core Web Vitals comparison
And if this pattern feels familiar, it should. We saw this exact movie before — it was called Core Web Vitals. The early messaging was soft. It's just a tiebreaker, don't panic — while they quietly baked it into every tool they own. The people who waited around for Google to officially confirm it mattered were already behind by the time that confirmation came. I don't want you in that group. That's really the whole reason I'm making this episode.
Looking beyond Google Search
And here's something else most people miss. Google Search saying it doesn't use llms.txt does not mean Claude isn't reading it. Or Perplexity. Or Copilot. If you only optimize for what Google publicly endorses, you're leaving the entire rest of the AI ecosystem sitting on the table.
Why llms.txt cannot fix a weak website
But — and this is the part that actually matters, so stay with me — llms.txt does not work in a vacuum. It never has.
A bare file, sitting on top of a thin, weak, badly built website, pointing an AI at your worst content, does nothing for you. You've handed over a beautiful table of contents for a book nobody should read.
The file is not a magic lever, and anyone selling it to you that way is selling you the keywords meta tag all over again.
Brand context, FAQs, services, and proof
It only earns its keep as part of a system. Picture an AI agent landing on your site with no llms.txt. It's like a new employee showing up on day one with no onboarding, no org chart — just a building full of filing cabinets. They'll figure out what you do eventually, but they're going to pull open a lot of wrong drawers first.
And when an agent has to wade through all that noise, two bad things happen. First, it hallucinates — confidently tells a customer something about your business that is flat wrong. Wrong hours, a product you don't sell, a policy you never wrote.
Second, the good information goes stale halfway through the task, and the answer curdles even though the AI was trying to get it right.
A clean, current llms.txt cuts down on both.
But the file itself is just the front door. What actually keeps the AI from getting you wrong is the real context behind that door — your brand information, your FAQs, your actual services, your proof. That's the difference between an AI that represents you accurately and one that makes things up about you to someone who's ready to buy.
Brian’s verdict and next steps
So when someone asks me if llms.txt works, here's where I land. The file by itself, floating in a vacuum? Overrated. The file as the front door to a site that's genuinely built for AI? That's quietly becoming essential — and most businesses haven't built it yet.
If you're sitting around waiting for Google to announce that it matters, you are going to be late. They are never going to hand you a clean announcement. They're going to build it into the tooling, ship it on their own sites, and let the audit reports do the talking. They've already shown their hand. You just have to watch their hands instead of their mouth.
Follow MAXVolume and connect with MAXPlaces
That's MAXVolume for today. If this was useful, do me a favor and follow the show, so the next one finds you. And if you want someone to look at where your own site actually stands with all of this, reach out to my team over at MAXPlaces. I'm Brian Winum. Thanks for turning it up. I'll see you next time.
