Obi Guard

Obi Guard builds the inspection layer for enterprises shipping LLMs and AI agents in production. We sit in front of every model call — inspecting prompts and completions, enforcing policy as code, and maintaining an append-only audit record that regulators actually trust.

Here's the problem we solve: traditional security doesn't work for AI. Organizations deploy large language models without visibility into what they say, what data they expose, or whether they follow policy. We fix that. Our gateway catches policy violations in real time, red-teams agents for bias and jailbreaks, and gives your security and legal teams an auditable record for every decision.

We're built by security veterans who've lived the problem. Our CEO spent 26 years in IT and telecom governance; our CTO has 15+ years in fintech and defensive coding. We're Malaysia-based, incorporated in December 2024, and already backed by Antler. We're hiring engineers and security researchers in Kuala Lumpur who want to solve a genuinely hard problem — not a demo audience.

Why this matters for Southeast Asia: organizations in Malaysia, Singapore, and Indonesia are shipping AI fast but without the governance framework that mature workloads require. We bring the same rigor you'd apply to data, identity, or infrastructure to every model call. Framework mapping to NIST AI RMF, ISO/IEC 42001, OWASP LLM Top 10, Malaysia's AISCF, and Singapore's MGF. Most customers go from first conversation to enforced policy in production within two weeks.