Optimize for AI · KXCO Cloud
AI crawlers are readingyour competitor's site.They're ignoring yours.
Most sites accidentally block the crawlers that feed AI answer engines — while trying to block the ones that train AI models. We fix the technical foundation. From $500.
Fix it from $500 →The problem
Not all AI crawlers are the same.
Most sites treat them as if they are.
There are two types of AI crawler. Training crawlers scrape your content to train language models. Retrieval crawlers index your site so AI answer engines — ChatGPT, Perplexity, Claude — can cite you in responses. One you may want to block. The other you need to allow. They have different User-agent strings. Most robots.txt configurations block both.
Training crawlers
You may want to block
Retrieval / search crawlers
Must allow to get cited
The common mistake: adding a blanket Disallow: / rule for GPTBot or PerplexityBot blocks both training and retrieval for that vendor. Your site disappears from ChatGPT Search and Perplexity answers. The fix is a two-line change per crawler — once you know which User-agent string is which.
What we deliver
Six concrete deliverables.
Every one of them documented.
AI Visibility Audit
We query ChatGPT, Claude, Perplexity, and Gemini with standardised questions about your category and brand — multiple queries per model, since AI answers are non-deterministic. You receive a plain-English report: what each model currently says about you, whether you are cited, and what is blocking citation.
llms.txt + Context Files
We write and deploy llms.txt at your domain root following the llmstxt.org specification. Immediate confirmed value: developer AI tools — Cursor, Claude.ai, ChatGPT — load this file when users ask about your product, making your docs and features more accurately described. It also positions you correctly as AI crawling formalises over time.
AI Crawler Configuration
The highest-value fix on this list. Training crawlers (GPTBot, Google-Extended, CCBot) and retrieval/search crawlers (OAI-SearchBot, PerplexityBot, Claude-SearchBot) are different bots with different User-agent strings. Most sites block everything with a single rule. We configure robots.txt so you control training data use while staying fully visible to AI answer engines.
Structured Data & JSON-LD
We write schema.org markup — Organization, FAQPage, Product, BreadcrumbList — for your key pages. Google has confirmed structured data improves inclusion in AI Overviews. It is also the most machine-readable signal you can give any AI that visits your site, regardless of which model.
Content Architecture Audit
We audit up to 10 key pages for AI extractability: heading hierarchy, claim-first paragraph structure, factual anchors, and table markup for comparison content. Deliverable: annotated recommendations or direct HTML patches — you or your developer implements them, or we push directly to your repo.
Citation Monitoring
We run a standardised query set across the major AI models weekly. Monthly report: citation rate per model, verbatim answer excerpts, and what changed month-over-month. Honest disclaimer: AI answers are probabilistic — we report trends and flag regressions, not guaranteed positions.
Proof of work
This is what you get.
Sample outputs from a Starter engagement. Your files will reflect your actual domain, product, and content.
# Deployed at: https://yourdomain.com/llms.txt # Spec: llmstxt.org — AI context file for developer tools # Your Company Name > One-line summary of what your product does. ## Products - [Product Name](https://yourdomain.com/product): What it does and who it is for. - [Second Product](https://yourdomain.com/product-2): What it does and who it is for. ## Documentation - [Getting Started](https://yourdomain.com/docs): Quick start guide. - [API Reference](https://yourdomain.com/api): Full API documentation. ## Optional # Lower-priority content — AI tools load this only if context allows - [Changelog](https://yourdomain.com/changelog): Recent changes and releases. - [Blog](https://yourdomain.com/blog): Technical articles and case studies.
Sample llms.txt — deployed at /llms.txt on your domain. Spec: llmstxt.org
Pricing
From $500. One-time or ongoing.
All work done by KXCO. No retainer required to start. Every engagement is documented and yours to keep.
Starter
$500
one-time
Get the foundation right.
Full Service
$1,500
one-time · recommended
Get the foundation right and get cited.
Ongoing
$300
/month
Stay visible as AI evolves.
Prices in USD. All deliverables are yours to keep regardless of future engagement. Questions? Talk to us →
FAQ
Technical questions, direct answers.
Does llms.txt actually improve my ranking in ChatGPT or Perplexity?+
Not in the way robots.txt affects Google rankings. As of mid-2026, no major AI company has confirmed they crawl llms.txt for search-time retrieval — Google's John Mueller stated directly that no AI system currently uses it for ranking. Its confirmed value is for developer AI tools: Cursor, Claude.ai, and ChatGPT load this file when a user asks about your product, making your documentation more accurately described. We also deploy it as forward-positioning: as AI crawling formalises, your site will be correctly structured before your competitors.
What if Perplexity ignores robots.txt anyway?+
Perplexity has two crawlers. PerplexityBot — their search indexer — respects robots.txt and is what determines whether Perplexity cites you in answers. Perplexity-User — their user-initiated fetcher — officially ignores robots.txt (Perplexity's own documentation confirms this). Our configuration targets PerplexityBot. Perplexity-User requires WAF IP-blocking if you want to control it, which is out of scope for this service.
How do I know if citation improves?+
The honest answer: you will see a trend, not a proof. AI citation is non-deterministic — the same query returns different answers on different days, and model updates can change citation patterns overnight. We establish a query baseline on day one, run the same set weekly, and report month-over-month trends. We do not claim guaranteed citation improvements; we report what is measurably changing.
What platforms and site types does this work with?+
Any site with a publicly accessible domain. For llms.txt and robots.txt, we need the ability to add files at the domain root — via repo access, FTP, or your hosting control panel. For JSON-LD, we deliver schema markup you add to your HTML head. For content architecture, we deliver annotated recommendations or HTML patches. We work with Next.js, React, WordPress, Webflow, static sites, or any CMS.
Do you need CMS admin access or repo access?+
For Starter: no. We deliver files and instructions; you or your developer deploys them. For Full Service: repo read access is helpful for the content audit, write access speeds up patching but is not required. We never need database credentials or admin passwords. For Ongoing: we need a way to be notified of new pages — a feed, a Slack message, or a weekly check-in call.
Why is this service on a post-quantum security platform?+
KXCO Cloud is built to fix technical infrastructure that is silently wrong. A misconfigured robots.txt that accidentally blocks AI retrieval crawlers is the same category of problem as a misconfigured TLS stack that leaves you on SHA-1: both are invisible, both have real consequences, and both require someone who knows what to look for. We fix the robots.txt the way we fix the crypto — precisely, with documentation, and with a record of what changed.
Get the technical foundation right.From $500.
The technical layer every GEO agency assumes you have — and never checks. We build it, document it, and hand it over.
Get started →All work done by KXCO · Deliverables are yours to keep