A few days ago an engineer in the US found my site through ChatGPT — searching for 1:1 agentic-AI training — and reached out. That wasn't luck. It's GEO: optimizing to be found and cited by AI.
SEO vs GEO
SEO gets you ranked in Google. GEO — Generative Engine Optimization — gets you quoted by ChatGPT, Claude, and Perplexity. These tools answer in prose and name a few sources. The goal shifts from "rank #1" to "be the source the model trusts enough to cite."
Why it matters: buyers now ask an AI "who can train my team on LangGraph?" — and act on the answer.
1. A plain-text profile for AI: llms.txt
LLMs love clean, quotable text. llms.txt is an emerging convention — a Markdown
profile at your root that crawlers can lift facts from without fighting your HTML.
# llms.txt — Arjun Thakur (arjunthakur.dev)
## Who is Arjun Thakur?
Principal AI Engineer, Fractional CTO, and founder of Yuvan.
Builds production agentic AI — LangGraph, RAG, WebRTC voice agents.
## Available for
- 1:1 & team AI training (agentic AI, LangGraph, RAG, MCP)
- AI engineering consulting and Fractional CTO
2. A connected entity graph (JSON-LD @id)
Don't scatter separate Person, Organization, and WebSite
blobs. Give each a stable @id and link them, so engines resolve you as one
entity — not three strangers with the same name.
{
"@type": "Person",
"@id": "https://arjunthakur.dev/#person",
"name": "Arjun Thakur",
"worksFor": { "@id": "https://arjunthakur.dev/#yuvan" },
"makesOffer": [
{ "@type": "Offer", "itemOffered":
{ "@type": "Service", "name": "AI Engineering Consulting" } }
]
}3. Answer questions in Q&A (FAQ schema)
LLMs extract question-answer pairs beautifully. Real FAQs, marked up as FAQPage,
become ready-made citations.
{
"@type": "Question",
"name": "Does Arjun offer 1:1 AI training?",
"acceptedAnswer": { "@type": "Answer",
"text": "Yes — private 1:1 mentoring in agentic AI,
LangGraph, RAG, and MCP, live online, worldwide." }
}4. Let the AI crawlers in
Most sites silently block them. I explicitly allow them in robots.txt:
User-agent: GPTBot
Allow: /
User-agent: ClaudeBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: CCBot # Common Crawl — feeds many models
Allow: /5. Get indexed fast
A fresh sitemap.xml plus IndexNow pings Bing and Yandex instantly —
and Bing's index feeds ChatGPT and Perplexity. Submit to Google Search Console for the rest.
The significance
AI answers are sourced from search indexes (Bing/Google) and Common Crawl. Content that is structured, quotable, and corroborated across your site and the web is what an LLM will confidently cite. GEO isn't a trick — it's making the truth about you easy for a machine to read. The proof: a real, paying lead that came straight from ChatGPT.
The site is open source
I've open-sourced this site — the schema, llms.txt, robots.txt, sitemap,
and IndexNow setup are all there to copy. Want the source code to replicate this on your own site?
Reach out and I'll share the repo.
I'm Arjun Thakur — Principal AI Engineer and Fractional CTO. I build production agentic AI (LangGraph, RAG, voice) and help teams do the same through consulting and 1:1 & team training. More at arjunthakur.dev.