Last week, I had the honour of speaking at LegalTechFest 2025 in Singapore — a key regional forum for legal professionals, technologists, and public-sector leaders to explore the future of law and regulation.

It was a proud moment for both myself and Thailand, as I was the only speaker from a Thai LegalTech company, representing our work with the Thai Government to modernize legal infrastructure and regulatory systems with our TH2OECD initiative.

Rather than showcasing technology, I shared lessons from over a year working at the intersection of law, language, and AI. Below are the key points I presented on stage — grounded in practice, not theory.

1. Diversity is the Core Challenge

Asia — and especially ASEAN — is home to diverse legal systems (Civil Law, Common Law, and hybrids), with no shared working language.

This legal and linguistic fragmentation impacts everything — from workflows to terminology. Tools that succeed in one market may not work in another. Localization isn’t enough; solutions must be designed for this complexity.

2. Civil Law Requires Textual Precision

In Civil Law systems like Thailand’s, interpretation comes from written statutes, not precedent.

AI must not invent or paraphrase legal clauses.

Every word must be traceable to the source law.

Even small errors undermine trust, especially in public-sector use.

This has shaped our system design — prioritizing accuracy, traceability, and human oversight.

3. The Devil is in the (Linguistic) Details

Legal meaning doesn’t always survive translation. Terms like “Goodwill” or “Reasonable” may carry very different meanings in Thai than in English.

Worse, Thai-language prompts often produce inconsistent outputs from models trained primarily in English. Language isn’t just a UX issue — it’s a technical and legal risk that must be addressed head-on.

4. New Human-AI Workflow is the Answer — Chat is Not Universal for Non-English Markets

Many LegalTech tools rely on chat-based interfaces, assuming users are fluent in English and skilled at prompting.

But in Thailand — and across much of Asia — users are often non-English speakers and non-experts. This makes prompt-based AI difficult to scale with trust.

In our TH2OECD project, we’ve used structured, guided workflows to deliver consistent results without relying on perfect prompts. This approach is more inclusive, scalable, and reliable.

5. ASEAN as a Global Testbed

ASEAN’s complexity isn’t just a challenge — it’s a proving ground. If LegalTech works here, it can scale globally.

Through TH2OECD, we’ve reduced a regulatory alignment process that normally takes 7–10 years into just 18 months — powered by AI-human collaboration. Early results suggest it could be a model for other emerging economies.

Final Thought

LegalTech in Thailand — and across ASEAN — is still early, but evolving fast. Success depends not just on good technology, but on respecting legal traditions, understanding language, and designing for real-world users.

I’m fortunate to work with visionary partners like the Office of the Council of State (OCS), especially Dr. Narun Popattanachai who believe in AI’s potential to reform Thailand’s legal system — and who trust me and the STelligence team to make that potential real.

Thank you to LegalTechFest 2025 for the opportunity to share these reflections. I hope our journey contributes to the broader conversation — not just about what LegalTech can do, but how it should be done.