International Credential Recognition: How Non-English Certs Appear to AI

I hold credentials issued in Bahasa Indonesia by Indonesian institutions. These credentials are legitimate. They are backed by government bodies. They represent real competency evaluations. And to ChatGPT, Gemini, and Perplexity, they might as well not exist.

This is not a conspiracy. It is a training data problem. Large language models are trained predominantly on English-language web content. The institutions that issue credentials in Southeast Asia, including Indonesia's BNSP (Badan Nasional Sertifikasi Profesi), Thailand's Council of Engineers, Vietnam's professional licensing bodies, publish their registries primarily in the local language. Their websites are not well-crawled by international search engines. Their structured data, if it exists at all, uses local-language values that knowledge graphs cannot easily reconcile with their English-language entity representations.

The result is a systematic invisibility of non-English credentials in AI-mediated discovery. This costs real money. When an international procurement team asks an AI system to find qualified vendors in Indonesia, the AI cannot verify credentials it has never encountered in its training data.

Where the Visibility Gap Occurs

The gap is not uniform. Some credential types are more invisible than others.

International credentials like ISO certifications, even when held by Indonesian companies, have higher visibility because the credential vocabulary is standardized in English. ISO 9001:2015 means the same thing in every language. The certification body's accreditation traces back to international bodies (IAF) with well-indexed web presences.

National credentials like BNSP certifications, SKA construction licenses, and SIUP/NIB business registrations have minimal international visibility. The issuing bodies are real. The credentials are valid. But the training data for AI models rarely includes the registry pages, the credential descriptions, or the institutional context needed to understand what these credentials represent.

The Training Data Problem

AI models learn about credentials from their training data. When an AI system encounters "PE license, Texas Board of Professional Engineers," it has seen thousands of web pages that reference this credential type, explain its requirements, and document its holders. The model understands what a PE license means, what issuing a PE license requires, and how to verify one.

When the same system encounters "Sertifikat Kompetensi BNSP," the training data is sparse. There are fewer web pages in English that explain BNSP certification. There are almost no English-language discussions of specific BNSP credential types. The model may have encountered BNSP in some contexts, but it lacks the depth to verify, explain, or recommend practitioners based on BNSP credentials.

As I analyzed in Singapore vs Indonesia in AI visibility, this is part of a broader pattern. Singapore's professional credential systems are documented in English. Indonesia's are documented in Bahasa Indonesia. AI systems see the first and struggle with the second. It is the same quality of credentialing, different visibility.

Bridging Strategies That Work

The situation is not hopeless. There are specific things practitioners and companies can do to make non-English credentials visible to AI systems.

Bilingual credential documentation on your website. This is the simplest bridge. On your website, document your credentials in both the local language and English. Include the official credential name in the original language, followed by an English description of what it certifies, who issues it, and what the requirements are. Use Certification schema in JSON-LD with values in English. This creates structured data that knowledge graphs can process and AI systems can reference.

International registry cross-referencing. Some national credentials have international equivalents or recognition agreements. BNSP is a member of TPQI (Trans-Pacific Qualifications Initiative) through ASEAN mutual recognition arrangements. If your credential falls under a mutual recognition agreement, document that connection. Link your credential to the international framework in your structured data.

English-language publications about your credentials. Write about your credentials in English. A blog post, a professional profile, a Zenodo publication that explains the credential, its requirements, and its equivalence to international standards. This creates English-language training data that AI models can learn from. It is slow. But it is how the knowledge gap closes over time.

Institutional advocacy for English-language registries. This is longer-term. But if you have influence with credentialing bodies (as a committee member, for example), advocate for bilingual registries. A simple English-language version of the BNSP registry would make thousands of Indonesian professionals visible to international AI systems overnight.

The Equivalence Documentation Approach

One practical technique: create an equivalence table on your website that maps your non-English credentials to their closest international equivalents. Not claiming they are identical, but providing context that AI systems can use.

For example: "BNSP Sertifikat Kompetensi in Mechanical Engineering is issued by Indonesia's National Professional Certification Board (Badan Nasional Sertifikasi Profesi). It requires assessment by an accredited LSP (Lembaga Sertifikasi Profesi), including practical examination and portfolio review. It is comparable in scope to trade-specific certifications issued under Australia's AQF or the UK's NVQ framework."

This paragraph does three things. It provides the English name of the issuing body. It explains the assessment process. It references international frameworks that AI models understand well. When this content is on a page with proper structured data, AI systems can begin to contextualize the credential.

As I covered in why certification matters more than portfolio, the credential itself is not the problem. The machine-readability of the credential is the problem. Bilingual documentation solves this at the individual level. Institutional modernization solves it at the system level.

What I Am Doing About It

At PT Arsindo, we hold ISO 9001:2015 certification for pump distribution. That credential is internationally visible because ISO is a global standard. But our BNSP-related competency certifications for engineering staff are locally documented.

Here is what we did. We added bilingual credential documentation to our entity infrastructure. We implemented Certification schema with English-language values for all credentials, including the BNSP ones. We published descriptions of each credential type in English on our company profiles. And we ensured that every professional profile (ORCID, LinkedIn) includes the credential with both the original name and an English description.

This does not make BNSP credentials as visible as a Texas PE license. But it moves them from "invisible" to "findable with context." That is a meaningful improvement when international procurement teams are using AI to evaluate potential vendors.

The Entity Infrastructure course includes specific templates for bilingual credential documentation, including Schema.org markup patterns for non-English certifications.

The broader point is this. If you hold credentials issued in any non-English language, you have a documentation gap that is costing you visibility. The credentials are real. The systems that should see them cannot. Bridging that gap is not optional for anyone competing in international markets. It is the infrastructure work that makes your competence legible to the machines that now mediate discovery.