Ask ChatGPT where to buy a genuine appliance in Kathmandu, and it answers confidently. What it won't tell you is that half the sites it could be pulling from never gave it permission to look.
I found this by accident, auditing a client's site earlier this year. Everything else checked out. Schema markup was in place, the content was clean, the pages loaded fast. Then I opened robots.txt, the small text file every website has that tells search engines and crawlers what they're allowed to touch. Buried in a block labeled "Cloudflare Managed Content" were five lines that stopped me:
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /
User-agent: Google-Extended
Disallow: /
The site wasn't blocking Google. It was blocking the crawlers behind ChatGPT, Claude, and Gemini's AI Overviews. Nobody had put those lines there on purpose. Cloudflare adds them by default to a lot of sites, quietly, as part of a "content signal" setting most people never open, let alone read. I've since checked this on several other Nepali business sites out of curiosity. It's not rare. It's close to the default.
What GEO actually means, in one sentence
Generative Engine Optimization (GEO) is the discipline of making sure AI systems can read, understand, and cite your content when someone asks them a question instead of typing it into Google. SEO gets you found in a list of ten blue links. GEO determines whether an AI assistant mentions you at all when a customer skips the list entirely and just asks.
The four crawlers that actually matter
Not every bot with "bot" in the name is worth worrying about. Four are:
OpenAI's crawler. Blocking it means your content can't be used to train future ChatGPT models, and depending on configuration, may also affect whether ChatGPT can browse and cite you live.
Anthropic's crawler. Performs the exact same indexing function for Claude, Artifacts, and conversational Claude integrations.
The one that surprises people. A separate crawler from regular Googlebot. Blocking it doesn't affect normal search rankings, but opts you out of Gemini and Google AI Overviews.
Belongs to Perplexity, which matters more every quarter as it grows into a real search alternative for direct, cited answers over lists of links.
Whether you should block these is a legitimate business decision. Some publishers deliberately keep AI companies from training on their content, and that's a defensible call when you make it on purpose. The problem I keep running into isn't businesses making that call. It's businesses who don't know the call got made for them, by a default setting, months before anyone on the team ever thought about AI search at all.
How to check your own site in two minutes
Type your domain followed by /robots.txt into your browser address bar: yourbusiness.com.np/robots.txt. Read what comes back.
Look for a section that starts with something like # BEGIN Cloudflare Managed content. Scan the lines under it for Disallow: / sitting beneath User-agent: GPTBot, User-agent: ClaudeBot, or User-agent: Google-Extended. If you find them and nobody on your team put them there on purpose, that's your answer, and it's been sitting there the whole time.
If your site runs through Cloudflare, the actual fix usually lives in the Cloudflare dashboard, under AI crawler or bot management settings, not in a file you hand-edit. That distinction matters, because I've watched the opposite mistake happen too: someone tries to "fix" this by pasting new Allow: / rules underneath the existing Disallow: / block instead of removing it. Now the file contradicts itself for the same five bots. Different AI crawlers resolve that conflict differently — some honor the first rule, some the last, and some decline to crawl a file that doesn't make internal sense. The fix is removing the block, not stacking a contradiction on top of it and hoping for the best.
A short checklist, if you'd rather just do it
- 01Open
yourdomain.com/robots.txtright now. - 02Search the page for "GPTBot," "ClaudeBot," and "Google-Extended."
- 03If any of them sit under a
Disallow: /you didn't write yourself, that's the leak. - 04If you're on Cloudflare, fix it in the dashboard's AI crawler settings, not by editing the text file directly.
- 05Re-check the URL after saving. The fix should be visible within minutes.
Why this matters more in Nepal specifically, not less
There's a version of this argument that only makes sense in markets with enormous search volume: AI search matters because millions of people already use it there. That's not really the Nepal case, and I don't think it needs to be.
The real case is smaller and sharper. Almost nobody here has fixed this yet. When I built Daewoo Nepal's SEO foundation from launch, checking and correctly configuring robots.txt for AI crawlers was one deliberate line item on the list, treated with the same care as any other technical SEO task, done once and done correctly. Two months in, the site is indexed, growing, and fully readable by every major AI crawler that might want to cite it. Most competitors selling the exact same categories haven't made that check at all, because most people don't know there's anything to look for.
That's the actual opportunity, and it's a narrow one worth taking seriously precisely because it's narrow. Not "AI search is huge in Nepal right now, get in early." It's simpler than that: almost nobody has checked whether their own site accidentally opted out of it, so checking is still a real advantage instead of table stakes everyone already cleared.
The fix costs nothing and takes an afternoon
No developer retainer. No new subscription. Open your robots.txt, read it like it's a decision someone made about your business (because whether anyone meant to or not, it is one), and fix the parts that were never actually a choice in the first place.
If you want a second pair of eyes on this or on the rest of your site's technical foundation, it's the kind of audit I do for a living.
