Want ChatGPT to Mention Your Shop? Start with LLMS.txt... or not yet

Want ChatGPT to mention your shop when people ask “where to buy a new laptop”? Meet  LLMs.txt - and the reality behind the hype

If you sell laptops, the dream is simple: someone asks an AI assistant where to buy, and your store gets mentioned (ideally with price, stock, and a link). A growing chorus says the new /llms.txt file is the key. Should you rush to add it?

Short answer: it’s worth knowing about, possibly worth adding (it’s cheap), but it won’t magically put you into AI answers today. Here’s the clear-eyed view.

1) What is llms.txt, and why is everyone talking about it?

llms.txt is a proposed, Markdown-formatted file you put at the root of your domain (e.g., example.com/llms.txt). Think of it as an AI-friendly index to your most useful pages, plus short descriptions, so language models can ingest context without fighting your site’s CSS/JS. It’s meant for inference-time use (helping models answer questions with your content), not a training opt-out or a crawl control like robots.txt.

How it compares:

• SEO / sitemaps: Built for search engines that follow links and rank pages. llms.txt is for models that summarize and ground answers. It doesn’t replace structured data, links, or feeds.

• robots.txt (RFC 9309): A decades-old standard crawlers use to decide what not to fetch. It’s about access control & crawl management, not content guidance. llms.txt doesn’t block or allow; it points to what’s important.

The idea originated with Jeremy Howard and Answer.AI in 2024, and many dev-tool docs adopted it quickly.

2) “Twitter has it” - does X actually use llms.txt?

There’s a lot of discussion on X (Twitter) about llms.txt, and several accounts track sites that add it. But as of September 22, 2025 we couldn’t verify that X itself exposes a public https://x.com/llms.txt. What does exist are high-profile examples from other big platforms, such as Stripe and Cloudflare, which publicly document their own llms.txt or llms-full.txt files for AI tools:

• Stripe added /llms.txt and Markdown versions of docs to make AI ingestion easier.

• Cloudflare docs publish llms-full.txt variants across product areas.

Directories and community posts on X track adoption (Cursor, Hugging Face docs, etc.), but that’s not the same as major AI search engines consuming the file.

3) We tested whether llms.txt is actually used: here’s a simple way (and what we saw)

The test (you can replicate in 15 minutes):

1. Publish a minimal llms.txt at yourdomain.com/llms.txt listing 3–5 high-value pages.

2. Ask AI assistants with browsing/grounding to answer questions that should obviously come from your pages (e.g., “What warranty does YourShop offer on Lenovo laptops?”).

3. Watch your server logs for a few days. You’re looking for any request to /llms.txt (and for the assistant fetching the actual linked pages).

4. Note whether the AI fetched only your homepage, or the specific docs/pages your llms.txt points to.

   
   

Our observation: in our checks this month, AI user-agents did not request /llms.txt; we saw occasional hits to the homepage or to individual pages, but no llms.txt lookups. That aligns with what many practitioners (and even Googlers) have reported publicly: major LLM providers haven’t committed to honoring llms.txt yet, and logs often show zero access.

There are also mixed claims in the wild—some directories and blogs say adoption is rising or that Anthropic ecosystems reference llms.txt in tooling—but even those sources generally don’t demonstrate systematic crawler use for AI search/ranking.

Bottom line from the test: Don’t expect adding llms.txt to immediately change what AI assistants say about your store.

Starter template you can copy:

# YourShop

> Laptops, accessories, and servicing in Zagreb. Shipping across EU. Support: support@yourshop.hr

## Policies
- [Warranty & Returns](/support/warranty) — Warranty terms, return windows, and process.
- [Shipping & Pickup](/support/shipping) — Delivery zones, fees, pickup options.

## Product Guides
- [How we grade refurbished laptops](/guides/grading) — A/B/C grades explained with photos.
- [Student laptop checklist](/guides/student) — RAM, SSD, CPU guidance for common budgets.

## Contact
- [Service & Repairs](/service) — Book a diagnostic or repair appointment.

4) So… will llms.txt help my shop get mentioned in ChatGPT?

Maybe later. Today, the signal that moves the needle most for “AI answers” is still classic, machine-readable commerce hygiene:

• Structured data on product pages (schema.org/Product) with price, availability, brand/GTIN, shipping, reviews.

• Merchant feeds (Google Merchant Center, Microsoft/Bing) so your catalog is in the ecosystems that AI assistants and answer engines frequently lean on.

• Clean, crawlable pages (no gated JS for specs/pricing), fast performance, and sensible robots.txt that doesn’t block product detail content or known AI user-agents if you want visibility.

As for llms.txt specifically, even industry trackers and SEO platforms say support is limited and evidence of impact is scarce right now—though it’s cheap to implement and may become a norm.

Verdict

• Add llms.txt if it’s easy. It won’t hurt, and it might future-proof your docs. Use it as a tidy index for AI tools (and your own agents) rather than a silver bullet for AI rankings.

• Don’t expect miracles. Our log-based test—and many others’—suggests major AI systems aren’t checking llms.txt during grounding.

• Do the fundamentals that do work today: high-quality product pages with correct structured data, inventory/pricing flowing to Merchant Centers, and a site that’s easy for machines to read. That’s what’s most likely to get your shop cited when an AI is asked where to buy a laptop.

If/when the big models officially adopt llms.txt, we’ll all cheer. Until then, treat it like bonus metadata, useful for tools and your own agents, while you double-down on the systems AI search engines already trust.