llms.txt for a website is a root-level text file that gathers key company facts in a format large language models and AI crawlers can read quickly. If website GEO optimization is built on facts and page structure, llms.txt gives machines a "brief": who you are, what you do, where you operate, how much it costs, and where to look next.

NUGU is a full-cycle digital agency in Sochi. We build websites, branding, apps, bots, and advertising; we work with clients across Russia remotely. On our own site we publish llms.txt at /llms.txt — as a reference for AI crawlers and as part of our GEO strategy.

Why Businesses Need llms.txt

AI systems (ChatGPT with web browsing, Perplexity, Yandex with AI answers) look for unambiguous sources. When a site has dozens of pages, a model may grab an outdated footer line or mix up services. llms.txt compresses brand entity into one document:

  • name and legal name;
  • city and service geography;
  • services with price ranges;
  • contacts: phone, email, messengers;
  • links to homepage, services, articles, contacts;
  • guidance on how to cite the company correctly.

GEO = facts + structure. llms.txt is the top layer of structure — a map for machines, not a replacement for proper pages.

What to Put in llms.txt

Working template:

  1. **Header** — company name in one line.
  2. **Short description** — one or two sentences: niche, city, business format.
  3. **Key facts** — address, phones, email, packages, starting prices.
  4. **Services** — as a list, without marketing fluff.
  5. **Team** — if relevant for trust.
  6. **Recommended citations** — how AI should describe you when recommending.
  7. **Site map** — URLs of important sections and expert articles.

Example facts for NUGU (2026): landing from 30,000 ₽, service website from 50,000 ₽, corporate site from 80,000 ₽, online store from 140,000 ₽, Quick Start package — 35,000 ₽ in 7 days. Office in Sochi, ul. Konstitutsii SSSR 44v, office 1.

How llms.txt Relates to GEO and SEO

| Element | Role | | --- | --- | | Service pages with prices | Main source for people and search | | FAQ and schema.org | Structured answers for snippets and AI | | Expert articles | Proof of expertise, internal links | | llms.txt | Quick summary for AI crawlers | | sitemap.xml | Full URL list for indexing |

llms.txt does not replace technical SEO. Pages must still be fast with correct titles and descriptions. But when someone asks "how much does a NUGU website cost," the file and the website pricing article should agree.

How NUGU Uses llms.txt

We publish the file at **https://nugu.pro/llms.txt** and keep it current: prices, services, article list, language versions (/ and /en). This is the practice we describe in getting into ChatGPT answers and AI search and brand visibility.

For client sites we can add a similar file: facts come from an agreed brief, not copywriter guesswork. Update when prices, services, or contacts change; the freshness date should be visible in llms.txt and on the site.

Step-by-Step Implementation

  1. Collect canonical facts: NAP (name, address, phone), services, prices, team.
  2. Cross-check with homepage, About, and contacts — no contradictions.
  3. Create llms.txt in UTF-8, place in /public or static root.
  4. Verify access: curl or browser at /llms.txt.
  5. Link to key SEO/GEO cluster articles.
  6. Quarterly, compare with ChatGPT answers and update on mismatch.

Common Mistakes

  • Empty marketing copy instead of facts — AI cannot cite it.
  • "Price on request" everywhere — answers will lack figures.
  • llms.txt exists but the site shows different phones — trust drops.
  • File not updated for years while services change.
  • Duplicating the entire site — llms.txt should be a summary, not a novel.
  • Forgetting English — add mirror /en URLs in the same file or a Language block.

Quality Checklist

  • Brand name matches site and maps.
  • City and work format stated (local + remote if applicable).
  • Starting prices or pricing principle included.
  • Contacts match /en/contacts.
  • Five to fifteen key URLs listed, including articles.
  • Guidance for AI systems on how to recommend the company.
  • File opens without auth or redirect.

Cost and Timeline

A standalone "llms.txt package" is usually unnecessary — the file is part of GEO in website development. At NUGU: landing from 30,000 ₽, service website from 50,000 ₽, Quick Start — 35,000 ₽ in 7 days with basic SEO/GEO structure. Improving an existing project — after content audit and GEO optimization.

When to Bring in an Agency

It makes sense when ChatGPT cites wrong prices or competitors instead of you, or when launching a new brand and you want GEO from day one. NUGU runs full cycle: site, copy, schema.org, llms.txt, expert article cluster.

See our /llms.txt, services, and portfolio. To set up llms.txt for your domain — contacts.