( Guide · AI search · Marketing )
How to get your business recommended by ChatGPT, Perplexity, Gemini and Google AI: a guide for Omani companies
By OGB Labs · Published
How AI assistants choose which businesses to mention
When someone asks ChatGPT, Perplexity, Claude or Gemini "which company builds apps in Oman" or "best clinic for physiotherapy in Al Khuwair", the assistant does not answer from memory alone. It runs web searches, fetches a small number of pages, and writes an answer with citations to them. Google calls this retrieval-augmented generation, or grounding, in its guide to optimizing for generative AI features: its AI features "rely on our core Search ranking systems to retrieve relevant, up-to-date web pages from our Search index", then review those pages to generate the response.
Google also describes query fan-out: the model generates "a set of concurrent, related queries" to fetch more results. A question about app development in Oman may fan out into searches about cost, timelines, Arabic support and payment gateways. A business whose pages answer those related questions appears in more of the fan-out results, and so more often in the final answer.
The other assistants work in a similar way, each with its own search index and crawler. OpenAI runs OAI-SearchBot to "surface websites in ChatGPT's search features", Perplexity runs PerplexityBot to surface and link sites in its results, and Anthropic runs Claude-SearchBot to improve search result quality for Claude users. The practical consequence is the same for all of them: if your pages are not crawlable, indexed and clearly written, there is nothing for the assistant to cite.
Step 1: let the right crawlers in
Each company publishes the user agents it uses and what each one does, so you can allow search crawlers while deciding separately about training crawlers. The table below is taken from the first-party documentation linked above. Allowing a crawler in robots.txt is not enough if a firewall or CDN blocks it: Perplexity and OpenAI both publish JSON files of their IP ranges so you can allow-list them, and both say robots.txt changes can take up to 24 hours to take effect.
- Decide on training separately from search. You can block GPTBot, ClaudeBot and Google-Extended and still be cited, as long as OAI-SearchBot, Claude-SearchBot and Googlebot are allowed. OpenAI notes that if you allow both OAI-SearchBot and GPTBot it may use a single crawl for both.
- Check your CDN and firewall. Cloudflare and similar services offer bot-blocking rules that stop these crawlers before they reach robots.txt. Perplexity's documentation walks through allow-listing by user agent and IP range for Cloudflare and AWS WAF.
- Watch your server logs after any change. If OAI-SearchBot or PerplexityBot never appears, something upstream is still blocking it.
- For a Next.js site, robots rules live in app/robots.ts; for WordPress, in the SEO plugin or a static robots.txt. Our own file names each crawler explicitly so the intent is unambiguous.
| Crawler token | Who | What it does | Block it and... |
|---|---|---|---|
| OAI-SearchBot | OpenAI | Finds pages to show in ChatGPT search answers | your site will not appear in ChatGPT search answers, only as a navigational link |
| GPTBot | OpenAI | Collects content that may be used to train OpenAI models | you opt out of training; ChatGPT search is unaffected |
| ChatGPT-User | OpenAI | Fetches a page when a user asks about it or uses a GPT Action; not a crawler | robots.txt may not apply, since the user triggered the fetch; it does not decide search inclusion |
| PerplexityBot | Perplexity | Surfaces and links sites in Perplexity results; not used for training | your site drops out of Perplexity results |
| Perplexity-User | Perplexity | Visits a page when a user's question calls for it | generally ignores robots.txt because a user requested the fetch |
| Claude-SearchBot | Anthropic | Indexes pages to improve Claude's search results | visibility in Claude's search answers may fall |
| Claude-User | Anthropic | Fetches a page when a Claude user's question calls for it | your content cannot be retrieved for user queries |
| ClaudeBot | Anthropic | Collects content that may contribute to training | you opt out of training |
| Googlebot | Crawls for Google Search, including AI Overviews and AI Mode | you disappear from Google Search and its AI features | |
| Google-Extended | Robots token only: controls use of your content for Gemini training and grounding in Gemini Apps and Vertex AI | no effect on Google Search inclusion or ranking, per Google |
Step 2: be indexed by Google and Bing with a snippet
Google is explicit about eligibility for its AI features. A page "must be indexed and eligible to be shown in Google Search with a snippet" (AI features and your website), and the site must be included in Search generative AI features in Search Console, which is the default. Google adds that meeting every requirement does not guarantee crawling, indexing or serving.
The same preview controls that limit classic snippets limit AI features. Google's May 2025 post on succeeding in AI search says: "More restrictive permissions will limit how your content is featured in our AI experiences." So check that no template adds nosnippet or a small max-snippet to pages you want cited.
Perplexity and ChatGPT search draw on their own indexes and on Bing, so verify the site in Bing Webmaster Tools as well as Google Search Console, submit the sitemap to both, and fix any pages that return errors. For a bilingual Omani site, make sure both the English and Arabic versions are indexed and linked with hreflang; the basics are in our SEO in Oman guide.
Step 3: publish answers a generic article cannot copy
Every assistant has many pages to choose from, so it favours pages with something the others lack. Google puts this bluntly in its guide: a first-hand review "provides a unique perspective based on personal experience, whereas a summary of existing content simply restates information already available elsewhere", and it contrasts commodity content like "7 Tips for First-Time Homebuyers" with content that "provides unique expert or experienced takes that go beyond common knowledge".
For an Omani business, the unique material is usually already in your head: what things cost in rials, how long approvals take, which areas of Muscat you serve, what goes wrong in local projects, and the questions customers ask on WhatsApp every week. Turn each recurring question into a page or section with a direct answer in the first paragraph, then the detail.
- Answer the question in the first paragraph, in plain language, then explain. Assistants quote paragraphs, and the first one is the easiest to quote.
- Use real local facts: OMR prices or ranges you actually charge, Omani rules with a link to the regulator, place names people use (Al Khuwair, Qurum, Al Mawaleh, Sohar, Salalah), and timelines from your own projects.
- Give each service its own page in English and Arabic. Arabic-speaking customers ask assistants in Arabic, and the assistants cite Arabic pages for Arabic questions.
- Use clear headings that match the questions. Google says people "appreciate it when web pages are organized by paragraphs and sections, along with headings"; it also says you do not need to chunk content into tiny pieces.
- Cite your sources for anything that is not your own experience: a ministry page, a law, a platform's documentation. Assistants are built to prefer verifiable claims, and unsupported statistics are the quickest way to be passed over.
- Add relevant photos and video of real work. Google's guidance says its AI features can bring in images and video, which gives a page more ways to appear.
- Do not generate dozens of near-duplicate pages for every variation of a query. Google says doing this to manipulate rankings or AI responses violates its scaled content abuse policy, and that its systems already understand synonyms.
Step 4: describe the business the same way everywhere
Assistants cross-check. If your website says you are a software company in Muscat, your Google Business Profile says you are a marketing agency, and a directory lists an old address in Ruwi, the model has three conflicting sources and is less likely to recommend you with confidence.
Write one sentence that says what you do, where you are and who you serve, and use it on the home page, the about page, your Google Business Profile, LinkedIn, Instagram, Oman Chamber of Commerce listings, supplier directories and partner pages. Keep the trade name, phone number and address identical, including the Arabic spelling.
Google's generative-AI guide specifically recommends keeping Google Business Profile and Merchant Center details current, because its AI responses "can include product listings, product information, and information about local businesses". Our guide to Google Business Profile in Oman covers verification and the fields that matter.
Earned mentions matter more than manufactured ones. Google warns against "seeking inauthentic mentions" and says its spam systems apply to AI features too. A real case study on a client's site, a trade association listing, a local news story or a supplier's partner page is worth more than a hundred directory submissions.
Step 5: structured data that matches the page
Structured data is not required for AI features; Google says there is "no special schema.org markup you need to add". It still helps, because it makes pages eligible for rich results and gives every reader of the page, human or machine, an unambiguous statement of your organisation, services, prices, opening hours and FAQs. Use Organization or LocalBusiness, Service, Product with Offer in OMR, FAQPage and Article, and keep the markup identical to the visible text: Google's guidelines require that "all the content in your markup is also visible on your web page".
What you can skip, according to Google
A whole industry of "AEO" and "GEO" advice has grown up around AI search. Google's position, in its optimisation guide, is that optimising for generative AI search "is optimizing for the search experience, and thus still SEO". It lists the tactics you can ignore for Google Search:
- llms.txt and other special files or Markdown: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search". Google adds that keeping one for other systems "will neither harm nor help" your Google visibility. We publish llms.txt and Markdown twins of our pages for assistants that may read them, and treat it as harmless, not as a ranking tactic.
- Chunking content into small pieces: "There's no requirement to break your content into tiny pieces for AI to better understand it."
- Rewriting pages in an "AI style" or stuffing long-tail keywords: AI systems understand synonyms and general meaning.
- Buying or seeding mentions across the web.
- Over-investing in structured data as if it were a requirement.
- Tools that claim to use "internal" Google metrics: Google says "No third-party tool has access to our internal ranking or AI systems."
How to check that it is working
Three sources tell you whether assistants are sending people to you. Google Analytics 4 now groups visits from ChatGPT, Gemini, Claude, Copilot and similar sources into an AI Assistant channel, and Search Console has a Generative AI performance report showing impressions in AI Overviews and AI Mode. Both are explained step by step in our guide to measuring AI search traffic in GA4 and Search Console.
The third check is manual: once a month, ask each assistant the questions your customers ask, in English and Arabic, and note whether you are mentioned and which page is cited. If a competitor is cited for a question you answer better, read their page and work out what the assistant found there that it did not find on yours. Google's advice on measuring value also applies: look beyond the click count at enquiries, WhatsApp conversations and sales, because it says visits from AI Overviews are "more likely to spend more time on the site".
A worked example: a Muscat maintenance company
Take an air-conditioning maintenance company in Muscat that wants to be recommended when someone asks an assistant "who services AC units in Al Hail" or, in Arabic, about annual AC maintenance contracts. The programme looks like this:
Fix access
Allow OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot in robots.txt and the CDN; confirm the site is indexed in Google Search Console and Bing Webmaster Tools with no snippet restrictions.
Publish one page per service in English and Arabic: annual contracts, emergency repair, duct cleaning, new installations, each opening with a direct answer and a price range in OMR, the areas covered and typical response times.
Add a page answering the questions the office hears every week: how often to service a split unit in Oman's climate, what a gas top-up costs, how to tell a compressor fault from a thermostat fault.
Align the Google Business Profile, Instagram bio, Oman Chamber listing and website so that the name, address, phone and one-line description match in both languages.
Earn three real mentions
A case study hosted by a property manager you serve, a listing with the relevant trade association, and a local supplier's authorised-partner page.
Measure monthly
The GA4 AI Assistant channel, the Search Console Generative AI report, and a manual check of the assistants in both languages.
How OGB Labs helps
We build websites in Muscat that are designed to be read by people and cited by machines: fast, bilingual, with crawler access set correctly, structured data that matches the page, and guides written from your real experience rather than generic copy. We also set up GA4 and Search Console so you can see the AI traffic arriving, and we run AI search optimisation (GEO) and SEO in Oman programmes for companies that want this done for them. See our website development and AI automation services, or message us on WhatsApp at +968 7111 0727.
( FAQ )
Questions people ask.
How does ChatGPT decide which companies to recommend?
When ChatGPT searches the web it uses OpenAI's OAI-SearchBot index and fetches pages that answer the question, then summarises them with links. Pages that are crawlable, indexed, clearly written and consistent with other sources about your business are more likely to be chosen. OpenAI says sites that block OAI-SearchBot do not appear in ChatGPT search answers.
Do I need an llms.txt file to appear in AI search?
Not for Google. Google states that it does not use llms.txt or other AI text files and that having one neither helps nor hurts Google visibility. Other assistants have not documented using it either. It is harmless to publish one for systems that may read it, but it is not a substitute for crawlable, well-written pages.
Which crawlers should I allow in robots.txt to appear in AI answers?
At minimum OAI-SearchBot (ChatGPT search), PerplexityBot (Perplexity), Claude-SearchBot (Claude) and Googlebot (Google Search, AI Overviews and AI Mode). GPTBot, ClaudeBot and Google-Extended control training and Gemini grounding and can be blocked without affecting search inclusion, according to each company's documentation.
Does blocking GPTBot stop my site appearing in ChatGPT?
No. OpenAI documents GPTBot as the training crawler and OAI-SearchBot as the search crawler, controlled independently. Blocking GPTBot opts you out of training; blocking OAI-SearchBot removes you from ChatGPT search answers.
Is GEO or AEO different from SEO?
Google says that from its perspective optimising for generative AI search is still SEO, because its AI features are built on its core ranking and quality systems. The work that matters is the same: crawlable pages, unique and helpful content, consistent business details and an honest reputation.
Do Arabic pages help an Omani business get cited by AI assistants?
Yes. Customers who ask assistants in Arabic are shown Arabic sources. Publish each service and guide in Arabic as well as English, link the two with hreflang, and keep the Arabic spelling of your name and address consistent across your profile and directories.
How long does it take to appear in AI answers?
Crawling and indexing changes take days; robots.txt changes take up to 24 hours to be noticed by OpenAI and Perplexity. Being chosen over competitors for a given question depends on your content and reputation and usually takes months of steady publishing, as with classic SEO.
Can I measure traffic from ChatGPT or Perplexity?
Yes. Google Analytics 4 assigns visits from recognised AI assistants the medium ai-assistant and groups them into an AI Assistant channel, and Search Console reports impressions in Google's AI Overviews and AI Mode. Our guide to measuring AI search traffic explains the setup.
Does Google's AI Mode use the same ranking as normal search?
Google says its generative AI features are rooted in its core Search ranking and quality systems and retrieve pages from the Search index, with additional fan-out queries. A page must be indexed and eligible for a snippet to appear, and there are no extra requirements.
( Related )
Service · 01
Websites
Company websites and web platforms designed around your brand, built to rank and to turn visitors into customers.
Service · 02
AI Automation
Agents and workflows that qualify leads, answer customers and move data between your tools — around the clock.
Service · 06
Digital Marketing
SEO, Google Ads, social media and AI search, run as one plan and reported against enquiries and sales.
Muscat · Oman
Website development company in Oman
Muscat · Oman
Generative engine optimization (GEO) for businesses in Oman
Muscat · Oman
AI specialists in Oman: AI agents, automation and custom AI for business
Muscat · Oman
Digital marketing agency in Oman
Muscat · Oman
SEO company in Oman
Muscat · Oman
Social media marketing agency in Oman
Guide · Analytics · AI search
Measure AI search traffic in GA4 and Search Console
Guide · Ecommerce · Saudi Arabia
Saudi VAT registration for online sellers
Guide · Software · UAE
UAE PASS integration for websites and apps
( Next step )
Want this done for you?
Tell us what you are building. We reply within one business day.

