Blog — 2026-08-06
By Emad Yahya — web developer & IT systems engineer, Dubai
No submission form, no paid placement and no advertising deal will put your business inside ChatGPT's recommendations. An assistant assembles its reply out of material it can find and cross-check on the open web, so getting recommended comes down to three conditions: being indexed — Google and Bing can crawl and understand your site; being corroborated — your name, services and location appear consistently in directories, reviews and articles the engines trust; and being quotable — your pages answer real questions in plain language an AI can lift into its reply. Meet those three conditions and you become a candidate. Skip them and you simply don't exist in the answer.
I'm Emad Yahya, a web developer and IT systems engineer based in Dubai. This is not theory for me: the AI-visibility SaaS I built puts a brand's market questions to the assistants themselves — ChatGPT-, Gemini- and Perplexity-style engines, with Google alongside them — and scores what comes back. I apply every fix on my own site, which ships as prerendered HTML with JSON-LD schema and an llms.txt file. Here is how the mechanics actually work, and what a Dubai business should do about them.
When someone asks an assistant for a recommendation — "best booking system developer in Dubai", "which clinic near Business Bay" — the assistant typically does two things. It draws on what its model learned during training, and, increasingly, it runs a live web search. ChatGPT's browsing has typically leaned on Bing's index, Gemini draws on Google's search, and Perplexity is built around live web retrieval.
The assistant then compresses what it finds — top-ranking pages, directory listings, "best of" articles, review profiles — into a short, confident reply. A search results page gives the reader ten links and lets them choose; the assistant hands back two or three names and stops. If yours is not among them, the customer never learns you were an option. That compression step is why GEO — generative engine optimization — is its own discipline: you are no longer competing for position eight on a results page, you are competing to be one of the few names the model considers safe to say.
Everything starts with the boring layer. If Google and Bing can't crawl, render and understand your site, no AI assistant will surface it, because the assistants sit downstream of those indexes. In practice that means fast pages, a clean sitemap, registration in both Google Search Console and Bing Webmaster Tools — Bing matters more than most Dubai businesses assume, precisely because of ChatGPT — and structured data.
Two details trip up modern sites. First, many crawlers still don't reliably execute JavaScript, so a site that only renders in the browser can look empty to the machines deciding whether to recommend you. My own site prerenders every page to static HTML for exactly this reason. Second, structured data (JSON-LD schema) tells engines what your pages are — a business, a service, an FAQ — instead of leaving them to guess. An llms.txt file, an emerging convention for handing AI assistants a curated map of your site, costs little to add; I publish one on this site, and my SaaS generates them.
An AI assistant is cautious about naming businesses. Before it recommends you, it typically wants your existence and your claims confirmed by sources it trusts: established directories, review platforms, comparison articles, an active Google Business profile. One self-hosted "we are the best" page is not evidence; the same facts repeated consistently across independent places is.
Consistency is the underrated part. Your business name, what you do and where you operate should read identically everywhere — website, directories, LinkedIn, reviews. Contradictions make you harder to verify, and models tend to skip names they can't verify. Note also which pages AI answers frequently draw from: "best X in Dubai" listicles and comparison pieces. Being genuinely present in credible ones — or publishing an honest, verifiable comparison yourself — puts you inside the exact source material the assistants quote.
Assistants assemble replies from fragments they can lift with confidence. Pages that bury the point under marketing copy give them nothing to lift. The fix is answer-shaped writing: one page per real question your customers ask, with a complete plain-language answer in the first two paragraphs, specifics instead of adjectives, and FAQ sections marked up with FAQPage schema so each question-and-answer pair is machine-readable.
This is where honesty becomes strategy rather than virtue. Concrete, verifiable claims — what you built, for whom, where it runs — survive AI compression. Hype does not.
This gap is why I built a product. You can track Google rankings, but AI answers are a black box — no console tells you what ChatGPT says about your market. So I built the AI Visibility & SEO/GEO Coach. It asks the assistants the questions your customers would ask — ChatGPT-, Gemini- and Perplexity-style engines, plus Google — and turns their replies into two numbers you can watch move: a visibility score, and how often your name gets mentioned at all. AI and Google are counted separately, so the gap between them is visible. What comes out is an ordered queue rather than a report, highest-impact fix first, and it generates the artefacts those fixes need — JSON-LD schema, llms.txt, robots.txt — so the work ships.
The loop matters more than the tool: measure, fix the top item, measure again. You can run it by hand — once a month, open the assistants, put your market's questions to them, and write down which names come back — as long as you keep the before and the after. One honest warning while you do: nobody, me included, controls a search index or a model's answers, so treat a promised ChatGPT mention as a sales line rather than a deliverable. The three conditions above are the part genuinely in your hands, and they are what moves the score. If you would rather hand the work over, that is what my SEO and GEO service in Dubai covers.
Because one of them could be verified and the other could not. The assistant is writing a short answer out of material it has just retrieved, so it favours a business whose site it could crawl and parse, whose details match what it found elsewhere, and whose pages hold a sentence it can lift without guessing. Fail any of those and you are not rejected — you never enter the shortlist in the first place.
Three labels, one target. GEO stands for generative engine optimization, AEO for answer engine optimization, and AI SEO is the informal name for both; each of them means getting your business named inside the answer an assistant like ChatGPT, Gemini or Perplexity writes for someone who asks about your market, instead of ranked in a list of links. The levers underneath are largely the ones classic SEO already uses: schema markup, pages that load and crawl cleanly, plain answers written where a reader can find them, and your brand mentioned consistently on sites that are not your own.
No — there is currently no way to buy a placement inside an AI assistant's organic recommendations, and anyone guaranteeing you a ChatGPT mention is selling. The answers are assembled from the open web, so the only durable route is making your business indexed, corroborated and quotable.
No, because the assistants sit on top of search rather than replacing it. When one of them needs live information it reaches into an index, and what it finds there is filtered by the same qualities Google rewards — speed, clean structure, authority. Most discovery in Dubai still happens in Google today, so treat GEO as a layer added on top of a technically sound site, not a substitute for one.
llms.txt is an emerging convention: a plain-text file at the root of your site that gives AI assistants a curated map of your pages and what they contain. It is not a magic ranking switch, but it costs little to add and makes your site easier for AI tools to read. I publish one on my own site, and my AI-visibility SaaS generates them.
SEO & AI Visibility (GEO) Services in Dubai · AI Visibility & SEO/GEO Coach — the SaaS · AI Developer in Dubai