By: Ace Zhuo, Founder of RankAISearch
Ask ChatGPT to recommend a good immigration lawyer in Sydney and it will answer with confidence. Ask it, in Bahasa Indonesia, to recommend a good notary in Surabaya, and something more revealing happens.
The answer gets vaguer. It leans on a couple of directory sites. It quietly defaults back to the English sources it trusts more. The assistant is not refusing you. It is guessing, and it is guessing from a worldview trained mostly on English text.
That gap is about to decide which brands win the next decade across Asia, and almost no one is treating it as the strategic problem it is.
Here is the shift underneath it. Buyers have started asking AI assistants the questions they used to type into Google: which clinic, which firm, which software, which supplier. A March 2026 survey of more than 1,000 B2B software buyers found that 71% now rely on AI chatbots to research vendors, up from 60% just seven months earlier, and 51% begin their buying process in a chatbot rather than a search engine (G2, The Answer Economy, April 2026).
Forrester’s 2026 Buyers’ Journey Survey, which collected responses from nearly 18,000 global business buyers, found that twice as many buyers named generative AI as their most meaningful research source than named any other source, outranking vendor websites and sales reps (Forrester, January 2026).
This is not a Western phenomenon that Asia will catch up to later. Asia-Pacific already leads the world in ChatGPT traffic share, and the fastest growth is here, with markets like Indonesia, the Philippines, South Korea, and Thailand among the quickest-rising (Microsoft AI Diffusion Report, January 2026; OpenAI Q1 2026 usage update). The demand has arrived. The problem is what happens when that demand meets the engines answering it.
Because those engines were built English-first. Around 40% of large language models today come from US-based companies, many trained predominantly on English, and at least one, OpenAI, has self-reported a Western and English-language bias (Carnegie Endowment for International Peace, January 2026). Independent evaluations keep finding the same pattern: models perform best in English and degrade in low-resource Asian languages, and English prompts often outperform the same question asked in Bangla, Hindi, or Urdu (arXiv, Better to Ask in English, 2024).
The result is a structural mismatch. Demand across Asia is increasingly multilingual and local. The supply of answers stays disproportionately English and foreign.
I run an AI search optimization firm based in Asia, and I see this mismatch from both ends every week. Western brands are slightly invisible to AI because most have not optimized for citation yet. Asian brands are far more invisible, because the engines were never well-fed on their languages, their local sources, or the way their customers actually phrase things.
Consider how people here really search. They rarely write clean, single-language queries. A buyer in Manila asks in Taglish, switching between Tagalog and English inside one sentence. A shopper in Kuala Lumpur mixes Malay and English without thinking about it. Someone in Bangkok or Ho Chi Minh City or Dhaka searches in a script and a vocabulary the training data barely covered. Researchers have specifically flagged that these models handle code-switching within and between Southeast Asian languages poorly (Carnegie Endowment, January 2026).
It gets more revealing. At least one study found that multilingual models tend to convert a non-English query into English internally, reason in English in their middle layers, then translate the answer back at the end (Carnegie Endowment, January 2026). So when a buyer asks in their own language, the system often answers an English version of their question, shaped by English sources, and hands it back translated. A local brand that deserves to be the answer never enters the room.
I want to be clear about what this is. It is an opening, and regional businesses should treat it as one.
The brands that fix this early will own their categories in AI answers for years. It works the same way the first companies to take SEO seriously in the 2000s built moats that took competitors a decade to erode. The AI-search land grab is moving faster and with far less competition, because in most Asian markets almost no one is optimizing for it yet. Even globally, only around 22% of marketers currently track AI visibility at all (Loganix synthesis of six studies, April 2026). The shelf is empty. Whoever stocks it first gets remembered.
So what does fixing it actually involve? Three things.
First, brands have to become legible to AI in their own market’s language, not only translated into it. There is a difference between having an English page and a Bahasa page, versus structuring your content so an answer engine can confidently extract, trust, and cite you when the question arrives in mixed or local language. Most “localized” sites in the region are translated brochures. Answer engines need structured, sourced, entity-clear information in the language of the query.
Second, local credibility signals have to exist where the models look. AI assistants lean heavily on third-party corroboration. In G2’s survey, 45% of buyers said citations from review sites were the single most confidence-inspiring signal in an AI answer. In Western markets that corroboration ecosystem is dense. In many Asian markets it is thin, which means the brands that deliberately build that footprint now face almost no competition for it.
Third, and this is the uncomfortable one for regional leaders, you cannot manage what you refuse to measure. Most businesses here have never once checked what ChatGPT or Gemini says about them when asked in their customers’ actual language. 69% of B2B buyers said they chose a different vendor than they had originally planned based on an AI chatbot’s guidance, and one in three bought from a vendor they had never heard of before (G2, April 2026). The decision is being made inside the answer, before anyone reaches your site.
The first step is embarrassingly simple. Run the queries your customers would run, in the languages they would use, and look at what comes back. It is usually a wake-up call.
There is a bigger stake here than any individual brand. As AI assistants become the default discovery layer, they also become an informal arbiter of which businesses, which sources, and which languages count as authoritative. If the region treats that as someone else’s problem, the default answers about Asian markets will keep being written from English sources and Western assumptions.
English-first AI is a temporary, fixable artifact of how these systems were built, not a permanent fact of life. That makes it the most valuable kind of problem a business can find, because the window does not stay open for long.
The brands that move while it is still open will not just rank in AI answers. They will be the reason the answer exists at all.
Ace Zhuo is the founder of RankAISearch, an AI search optimization firm working with brands across Asia, Australia, and the US to earn visibility in AI-generated answers.
The AI Search Blind Spot That Will Decide Asia’s Next Decade of Brands
By: Ace Zhuo, Founder of RankAISearch
Ask ChatGPT to recommend a good immigration lawyer in Sydney and it will answer with confidence. Ask it, in Bahasa Indonesia, to recommend a good notary in Surabaya, and something more revealing happens.
The answer gets vaguer. It leans on a couple of directory sites. It quietly defaults back to the English sources it trusts more. The assistant is not refusing you. It is guessing, and it is guessing from a worldview trained mostly on English text.
That gap is about to decide which brands win the next decade across Asia, and almost no one is treating it as the strategic problem it is.
Here is the shift underneath it. Buyers have started asking AI assistants the questions they used to type into Google: which clinic, which firm, which software, which supplier. A March 2026 survey of more than 1,000 B2B software buyers found that 71% now rely on AI chatbots to research vendors, up from 60% just seven months earlier, and 51% begin their buying process in a chatbot rather than a search engine (G2, The Answer Economy, April 2026).
Forrester’s 2026 Buyers’ Journey Survey, which collected responses from nearly 18,000 global business buyers, found that twice as many buyers named generative AI as their most meaningful research source than named any other source, outranking vendor websites and sales reps (Forrester, January 2026).
This is not a Western phenomenon that Asia will catch up to later. Asia-Pacific already leads the world in ChatGPT traffic share, and the fastest growth is here, with markets like Indonesia, the Philippines, South Korea, and Thailand among the quickest-rising (Microsoft AI Diffusion Report, January 2026; OpenAI Q1 2026 usage update). The demand has arrived. The problem is what happens when that demand meets the engines answering it.
Because those engines were built English-first. Around 40% of large language models today come from US-based companies, many trained predominantly on English, and at least one, OpenAI, has self-reported a Western and English-language bias (Carnegie Endowment for International Peace, January 2026). Independent evaluations keep finding the same pattern: models perform best in English and degrade in low-resource Asian languages, and English prompts often outperform the same question asked in Bangla, Hindi, or Urdu (arXiv, Better to Ask in English, 2024).
The result is a structural mismatch. Demand across Asia is increasingly multilingual and local. The supply of answers stays disproportionately English and foreign.
I run an AI search optimization firm based in Asia, and I see this mismatch from both ends every week. Western brands are slightly invisible to AI because most have not optimized for citation yet. Asian brands are far more invisible, because the engines were never well-fed on their languages, their local sources, or the way their customers actually phrase things.
Consider how people here really search. They rarely write clean, single-language queries. A buyer in Manila asks in Taglish, switching between Tagalog and English inside one sentence. A shopper in Kuala Lumpur mixes Malay and English without thinking about it. Someone in Bangkok or Ho Chi Minh City or Dhaka searches in a script and a vocabulary the training data barely covered. Researchers have specifically flagged that these models handle code-switching within and between Southeast Asian languages poorly (Carnegie Endowment, January 2026).
It gets more revealing. At least one study found that multilingual models tend to convert a non-English query into English internally, reason in English in their middle layers, then translate the answer back at the end (Carnegie Endowment, January 2026). So when a buyer asks in their own language, the system often answers an English version of their question, shaped by English sources, and hands it back translated. A local brand that deserves to be the answer never enters the room.
I want to be clear about what this is. It is an opening, and regional businesses should treat it as one.
The brands that fix this early will own their categories in AI answers for years. It works the same way the first companies to take SEO seriously in the 2000s built moats that took competitors a decade to erode. The AI-search land grab is moving faster and with far less competition, because in most Asian markets almost no one is optimizing for it yet. Even globally, only around 22% of marketers currently track AI visibility at all (Loganix synthesis of six studies, April 2026). The shelf is empty. Whoever stocks it first gets remembered.
So what does fixing it actually involve? Three things.
First, brands have to become legible to AI in their own market’s language, not only translated into it. There is a difference between having an English page and a Bahasa page, versus structuring your content so an answer engine can confidently extract, trust, and cite you when the question arrives in mixed or local language. Most “localized” sites in the region are translated brochures. Answer engines need structured, sourced, entity-clear information in the language of the query.
Second, local credibility signals have to exist where the models look. AI assistants lean heavily on third-party corroboration. In G2’s survey, 45% of buyers said citations from review sites were the single most confidence-inspiring signal in an AI answer. In Western markets that corroboration ecosystem is dense. In many Asian markets it is thin, which means the brands that deliberately build that footprint now face almost no competition for it.
Third, and this is the uncomfortable one for regional leaders, you cannot manage what you refuse to measure. Most businesses here have never once checked what ChatGPT or Gemini says about them when asked in their customers’ actual language. 69% of B2B buyers said they chose a different vendor than they had originally planned based on an AI chatbot’s guidance, and one in three bought from a vendor they had never heard of before (G2, April 2026). The decision is being made inside the answer, before anyone reaches your site.
The first step is embarrassingly simple. Run the queries your customers would run, in the languages they would use, and look at what comes back. It is usually a wake-up call.
There is a bigger stake here than any individual brand. As AI assistants become the default discovery layer, they also become an informal arbiter of which businesses, which sources, and which languages count as authoritative. If the region treats that as someone else’s problem, the default answers about Asian markets will keep being written from English sources and Western assumptions.
English-first AI is a temporary, fixable artifact of how these systems were built, not a permanent fact of life. That makes it the most valuable kind of problem a business can find, because the window does not stay open for long.
The brands that move while it is still open will not just rank in AI answers. They will be the reason the answer exists at all.
Ace Zhuo is the founder of RankAISearch, an AI search optimization firm working with brands across Asia, Australia, and the US to earn visibility in AI-generated answers.
Table of Contents