Defensive Intelligence
Defensive
Intelligence
Built by ninthLABS Ventures

About OBEL™

Defensive Intelligence
for a dangerous world

OBEL™ is the AI governance platform that lets regulated organisations use frontier AI safely - without compromising classified data, personal information, or compliance posture.

The name

OBEL™ - derived from the Obelisk, the ancient symbol of a monolithic pillar marking a protected boundary.

Technically: Operational Boundary-Enforced LLM.

Our Mission

Make AI safe for the work that matters most

Enterprises and government departments want to adopt AI. The problem is not the technology - it is the governance gap. PII leaving your network in prompts. Agents acting on real systems without oversight. Regulators asking for audit trails that do not exist. Developers running frontier AI with the full codebase in context and no organisational visibility.

OBEL sits between every user, agent, and tool and the model they reach - classifying, scrubbing, governing, and logging every interaction before inference. Governed chat, autonomous agents, document generation, knowledge bases, and developer CLI tools all through the same pipeline. The result is confident AI adoption, not cautious AI avoidance.

58Compliance controls mapped across EU AI Act, NIST AI RMF, SOC 2 Type II, and ISO 27001
17MITRE ATT&CK techniques classified on every agent tool call
100%Of interactions governed before any model call - chat, agents, documents, CLI tools
0Fail-open states - every control gate is fail-shut by architecture, not policy

Our Story

Built from the problem, not the technology

2022

ninthLABS Ventures founded in Sydney, focused on AI, security, and automation for regulated industries.

2023

OBEL™ concept developed following observed PII and data leakage incidents in enterprise AI adoption. The question that shaped the product: can an organisation use AI productively without ever letting sensitive data reach a model?

2024

OBEL™ v0.1 deployed in closed beta with government and financial services pilot customers. ARGUS-i™ sovereign classification engine designed and validated against PSPF, ISM, and Five Eyes frameworks.

Apr 2026

OBEL™ v1.0 "Sovereign Lockdown" declared stable baseline. Multi-tenant, multi-provider, production-ready. ARGUS-i, PII scrubber, sovereign gate, and audit vault all declared immutable core components.

Jun 2026

Intelligence Hub launched with Decision Trace Ledger, Agentic APM, MITRE ATT&CK classification, Model Arena, and Prompt Analytics. Developer CLI proxy governance released for Claude Code, Aider, Gemini CLI, and Goose. SOC 2 Type II and ISO 27001:2022 compliance mapping completed. SIEM/CEF connector shipped.

Core Principles

What guides every decision

Governance in the pipe, not the policy

Controls are built into the architecture - not bolted on, not dependent on user compliance, not configurable out of existence. The scrubber, classifier, sovereign gate, and audit writer run on every request. There is nothing to remember and nothing to bypass.

Fail-shut by architecture

When a control fails, the system denies the request. OBEL never degrades to a pass-through. In regulated environments a silent failure is a compliance breach. Fail-open is not a trade-off we make under any circumstances.

Audit evidence, not audit theatre

The governance record is written before inference, not reconstructed after. An audit trail that is assembled from logs after the fact is not an audit trail. The Decision Trace Ledger exists because regulators are starting to know the difference.

Confident adoption, not cautious avoidance

The goal is not to slow AI adoption down. It is to remove the governance blockers that are slowing it down already. Boards, legal, security, and compliance teams should be able to approve AI deployment because the infrastructure is provably governed - not because they have run out of objections.

Vendor-agnostic by conviction

Locking organisations into a single model provider is a governance risk, not a governance solution. OBEL governs any provider through the same pipeline. The June 2026 Fable/Mythos suspension proved the point. OBEL customers kept running.

Shipped, not promised

Intelligence Hub, Developer CLI proxy, SIEM/CEF connector, MITRE ATT&CK classification, LLM-as-judge evaluation, red-team adversarial testing, and four-framework compliance mapping all shipped in 2026. The roadmap is real because the baseline is stable.

The Company Behind OBEL™

ninthLABS Ventures Pty Ltd

ninthLABS Ventures is an enterprise technology venture studio headquartered in Singapore with operations across Australia. The studio builds category-defining platforms at the intersection of AI, cybersecurity, and compliance - markets where deep domain expertise creates a structural advantage that pure technology teams cannot replicate.

ninthLABS draws on the operational network and domain expertise built across NINTH EAST Partners - a cybersecurity and professional services advisory firm with practices across Australia, New Zealand, Singapore, and Hong Kong. This network provides enterprise and government relationships and distribution reach that most early-stage software companies spend years building.

OBEL is the flagship platform - built to govern the AI governance gap at the moment the market is being legislated into mandatory compliance. It is production-ready, App Store approved, and serving customers across Individual Pro, Enterprise, and Government tiers. The Intelligence Hub, Developer CLI proxy, SIEM connector, and full compliance framework mapping shipped in 2026 alongside the stable v3.0 platform baseline.

Global Offices

HQ🇦🇺

Sydney

Barangaroo, NSW 2000

🇸🇬

Singapore

Marina Bay

🇺🇸

New York

Manhattan, NY

🇺🇸

San Francisco

SoMa, CA