I was scrolling Instagram and saw a marketing creator's Reel about ranking in Google without paying for ads. The hook was slick: "I don't pay a dime" with a screenshot of their business sitting in the local map pack.
The Reel said you need three things to get indexed and rank better: an XML sitemap, llms.txt, and markdown mirrors.
Sounded interesting. So I did what I do with most marketing claims now. I sent it to my AI coworker Viktor and asked if it was legit.
The Reel
I dropped the Instagram link into our Slack DM. Viktor pulled the thumbnail and identified the content immediately, even though IG blocks full video access behind login.
The Honest Breakdown
I told Viktor what the Reel was claiming and asked a simple question: do you agree, and do you know how to do all three?
This is where it got interesting. Viktor didn't just say "yes, let's do it." It broke down the claim piece by piece and told me which parts were real and which parts were oversold.
The short version:
- XML sitemap is the only one that actually affects Google indexing. I already had one submitted to Google Search Console. Table stakes. Done.
- llms.txt and markdown mirrors are NOT Google ranking factors. Google's own search team has publicly said they don't use llms.txt. There's no evidence markdown mirrors change traditional rankings either.
- Where they DO matter: getting cited in AI answers. ChatGPT, Perplexity, Google AI Overviews. They give LLMs a clean, curated version of your site to pull from.
So the Reel was packaging an AI visibility play as an SEO play. Not wrong exactly, just oversold.
Viktor's honest verdict: worth doing if the goal is showing up in AI answers. Won't move your Google rankings.
That distinction matters. If I'd just followed the Reel's advice blindly, I'd be expecting a Google rankings bump that was never coming. Instead, I knew exactly what I was getting and why.
From Fact-Check to Execution
The part that still surprises me about working with an AI coworker: the gap between "should we do this?" and "it's done" can be minutes.
I said "go ahead and execute." Viktor mapped my entire site, built a curated llms.txt index and 8 markdown mirrors, and deployed everything. Total elapsed time: about 16 minutes.
It also found something I didn't expect. Shopify was already auto-serving a /llms.txt and /agents.md at my site's root, but filled with generic "buy products via Shop Pay" boilerplate. Useless for a services site with zero products. Viktor overrode both with real, curated content.
Then I asked Viktor to do the same for my other business, ThreadPoint. Same thing: mapped the site (50+ blog posts, 11 service pages, 6 industry pages), built 20 markdown mirrors, deployed. Both sites covered in under an hour total.
What the Live Output Looks Like
Here's the actual llms.txt file Viktor built, live on my site right now. Every service, every page, every blog post, each with a link to the original HTML page plus a clean markdown mirror for AI crawlers.
You can see it yourself at chiefmarketingdad.com/llms.txt. That's the file AI answer engines like ChatGPT and Perplexity read when they want to understand what my site is about.
I wrote a deeper technical walkthrough of exactly what llms.txt and markdown mirrors are, how they work on Shopify, and why they matter for AI search visibility in a separate post here.
Why This Is How I Use AI Now
This whole thing took maybe 20 minutes of my time. I saw a Reel, sent a link, got an honest answer, said "go," and it was done across two websites.
But the valuable part wasn't the execution speed. It was the fact-check.
Marketing is full of Reels and threads and posts telling you to do stuff. Most of it blends real advice with hype. The creator in this Reel wasn't lying, but they were overselling (llms.txt helps with AI visibility, not Google rankings). That's a meaningful difference if you're deciding where to spend your time.
Having an AI coworker that will push back and say "this part is real, this part is oversold" before doing anything is worth more than the execution itself.
If you want to try this yourself, Viktor gives you $100 in free credits plus $50 off your first purchase ($150 total value, no credit card required). It runs right in Slack, and yes, it will tell you when a marketing hot take is only half right.
More real stories of me working with Viktor:
