You know whether your name comes up when customers ask AI, whether the facts are right, and how many visits AI sends each month, without relying on anyone’s promise of a top spot in ChatGPT.
Back to AI Web Development
Owners & marketing teams · 07 / 13
AI search visibility (GEO & AEO)
See what ChatGPT, Gemini, Perplexity and Google AI Overviews tell customers about your business, then fix it at the source: prices, branches and hours as clear text, in Thai and English.
Owners & marketing teams “I asked ChatGPT for the best aesthetic clinic in Bangkok and we weren’t on the list. When I asked about us by name, it quoted last year’s promotion and a branch we closed in January.”
The problem
Most owners find out by asking for themselves. ChatGPT names five clinics and yours is missing. Google’s AI Overview gives last year’s opening hours. Ask who makes a certain part in Thailand and a competitor comes up. The cause is usually the site. Prices and promotions sit in poster images and specs in PDFs. Some pages show their content only once the browser runs JavaScript, which Google itself notes some bots cannot do. The English pages carry half the facts of the Thai ones, and the website, the Google Business Profile and the Facebook page disagree on opening hours. When the facts contradict each other, an assistant may repeat the old one.
Nobody is measuring it either. GA4 now has an AI Assistant channel, but it counts only visits whose source is identified, and clicks from AI Overviews and AI Mode sit inside Organic Search with ordinary results. So marketing cannot say how many people AI sends, or what it is getting wrong, until a customer messages about a promotion that has ended or drives to a branch that has closed.
How we solve it
The work is often called GEO (generative engine optimisation) or AEO (answer engine optimisation). We start by collecting the questions customers really ask, from your LINE admins, your sales team and the search terms in Search Console, in Thai and English. We put them to ChatGPT, Gemini, Perplexity and Copilot, check Google’s AI Overviews and AI Mode, and record every answer: whether you appear, what is wrong and which sites are cited. That report sets the order of work. We then rewrite those pages so each question gets a clear answer in the first paragraph, with prices, branches, hours and conditions as text on the page in both languages, rendered on the server and quick to load. The Schema.org markup matches what the page shows, and your Google Business Profile is corrected to match the site.
We set crawler access in robots.txt with you, one crawler at a time, because each does a different job. OpenAI’s OAI-SearchBot builds ChatGPT’s search results, and GPTBot collects content to train its models. Anthropic separates Claude-SearchBot and ClaudeBot the same way. PerplexityBot builds Perplexity’s search results. At Google, AI Overviews and AI Mode use the same Googlebot as Search, while Google-Extended covers both training Gemini and using your content to ground answers in the Gemini app. So you can keep search open and close training, knowing what each choice costs. We add an llms.txt too. It is still a proposal that AI companies may or may not read, and Google says no such file is needed to appear in AI Overviews. Businesses with a catalogue or bookings get a feed or an MCP endpoint that agents can query for prices, stock or open slots. We run the same on our own site.
Nobody can guarantee that an assistant will mention you. The answers come from each company’s models and change over time. What we control is that your facts are clear, current, machine-readable and the same everywhere. Each month we re-run the same questions, compare them with the month before, and put the results on one dashboard with AI Assistant visits from GA4, figures from Search Console and Copilot citations from Bing Webmaster Tools. When an answer is wrong, the system drafts the correction. Prices, anything about treatment and any terms with legal weight are approved by your people before they go live.
How it runs
Work comes in from
- Real customer questions, Thai and English
- Existing pages, PDFs and images
- Prices, branches and hours from your systems
- Search Console and GA4
What the AI does
- Ask each assistant and record the answer
- Find facts that are wrong or missing
- Draft new pages and schema
- Re-check and compare each month
Where it lands
- Pages that answer the question
- robots.txt, llms.txt and feeds
- Dashboard of visits from AI
- Monthly report to the owner
Before and after
Before
After
No idea what AI tells customers about the business
A monthly record of what each assistant says, in Thai and English
Prices and branches sit in poster images crawlers can’t read as text
Prices, branches and hours are text on the page, the same in both languages
Nobody looks at visits from AI assistants on their own
A dashboard that separates visits from ChatGPT, Perplexity, Gemini and Copilot
What you get
- 01
An audit of what ChatGPT, Gemini, Perplexity, Copilot and AI Overviews say to your customers’ real questions, in Thai and English: whether you appear, what is said, which sources are cited
- 02
Pages rewritten so each question gets a clear answer, with prices, branches and hours as text in Thai and English, on fast pages rendered on the server
- 03
Schema.org markup that matches what the page shows (Organization, LocalBusiness, Product, FAQ where it fits), robots.txt rules set crawler by crawler, and an llms.txt file
- 04
A product feed or an MCP endpoint for businesses with a catalogue or bookings, so AI agents can ask your live system for prices, stock or open slots
- 05
The same questions re-run every month, and a dashboard that separates visits from ChatGPT, Perplexity, Gemini and Copilot from every other channel
Who gets what
No re-platforming: we work in your existing CMS and add server rendering only where a page needs it. robots.txt uses the crawler names from OpenAI, Google, Anthropic and Perplexity’s documentation, the MCP endpoint is read-only, rate-limited and exposes no customer data, and GA4 follows the site’s PDPA cookie consent.
Marketing stops asking ChatGPT one question at a time and pasting screenshots for the boss. Admins see fewer customers arriving with an expired promotion, because the site holds one current version of every price.
Who this fits
Brands & retailersClinics & hospitalsSchools & tutoringB2B manufacturers & distributorsHotels & resortsReal estate projects
Connects with what you already run
Google Search ConsoleBing Webmaster ToolsGoogle Analytics 4Google Business ProfileWordPress or existing CMSBooking system or ERP
Development process
- 1
Discover
Requirements, users and success metrics, with scope and price fixed before we start.
- 2
Design
UX and system architecture; the prototype is approved before anything is built.
- 3
Build
AI-accelerated sprints with a demo every week, reviewed by senior engineers.
- 4
Test
QA, security and performance verified against the agreed scope.
- 5
Launch & care
Production deploy, team training, and a monthly care plan.
Turn your business problem into a system that works for you
Tell us today — get an executive-ready proposal with the plan and budget.
Talk to an engineer about thisMore solutions in AI Web Development
See the whole service Companies & factories Multilingual corporate website Retail & brands Online store on your own domain Hotels & resorts Booking engine on the hotel’s own site Service firms & B2B Customer portal Marketing teams & agencies Campaign & landing pages Operations & IT Internal web app Clinics & aesthetics Clinic website with online booking Schools & universities Admissions site and student portal Real estate Project site with viewing appointments Restaurants & cafés Restaurant site with ordering and table booking Factories & wholesalers B2B catalogue and quote requests Tutoring schools & trainers Online course website
