Start a conversation

Make AI‑driven decisions explainable, governed, and defensible.

Governance, evidence, and automation systems for organisations where the cost of being wrong is high.

Enforceable governance Traceable evidence Auditable workflows

The accountability gap

If AI is influencing outcomes in your organisation, can you explain what it did, why, and how — to a regulator, or your board?

Most organisations can't. The governance was built for humans using software, not AI agents using agents.

Next step

A 20-minute discovery call to confirm your problem, assess fit, and outline what the system would look like.

Book the discovery call

What I Build

Operational systems for organisations where the cost of being wrong is high.

Data + Search

Evidence Intelligence

Turn thousands of emails, contracts, regulatory letters, and financial records into a searchable evidence layer that reconstructs timelines, maps causality, and prepares complex complaints and disputes. Every assertion traced to source document, page, and exact wording — built so AI agents and humans can reason across your evidence.

SQLite · PostgreSQL + RLS · ChromaDB · OCR pipelines

1,000+ Docs indexed
Timeline Reconstruction
Complaint Ready

Governance + Enforcement

Governance Architecture

Policy enforcement that AI agents actually obey. Seven atomic controls — purpose, data class, model version, human review triggers, blocking rules, audit logging, retention — enforced at the database layer via PostgreSQL row-level security, not in application code.

7 Primitives · RLS · JSONB decision traces · accountability binding

7 Primitives
RLS Enforced
Immutable Audit trail

Automation + Integration

Agentic Workflows

Transform manual processes into AI-native workflows where agents do the work and humans provide judgment. Connect M365, Salesforce, databases, and APIs through orchestration layers. Deploy reusable AI skill libraries with multi-lens analytical protocols so teams get consistent, repeatable analysis instead of ad-hoc prompting.

Claude Code skills · N8N · MCP servers · 7-lens protocol · CLI/API tools

33+ AI skills
6 Domains
Portable Protocol

Evidence + Regulatory

Compliance Intelligence

Forensic-quality evidence management for regulatory complaints, litigation support, and compliance investigations. Every assertion traced to source document, page, and verbatim quote. Chain-of-custody document processing from ingestion to submission-ready evidence packs across financial services, privacy, consumer protection, and administrative review.

Regulatory mapping · causal chain analysis · forensic language audit

Chain Of custody
Verified Cross-reference
Submission Ready

How It Works

The same architecture, applied to every domain. The methodology is portable — the infrastructure scales to need.

Intake

Docs, emails, PDFs, records, regulatory instruments

Data Layer

SQLite / PostgreSQL + FTS + ChromaDB semantic index

Analysis

Claude skills, 7-lens protocol, causal chain mapping

Quality Gate

Forensic language audit, blind-spot check, compliance review

Output

Defensible outputs, filed submissions, auditable decisions

Infrastructure

SQLite for edge deployments. PostgreSQL + RLS for production. Docker for development. Supabase or Azure for hosted. The intelligence layer is infrastructure-agnostic — designed so you're not locked in.

Why This Matters Now

Three things changed — and they changed what 'ready' looks like.

Software costs collapsed

AI agents now build and operate applications autonomously. Per-seat SaaS is dying; the application layer is commoditised. What still has value is your data, and the accountability infrastructure that lets you defend decisions made with it.

The bottleneck moved

The hard problem is no longer "how do we build this?" It's "can we specify what we need precisely enough that autonomous systems deliver it safely — and can we prove they did?" The skill that matters now is translating business intent into machine-enforceable constraints.

Tools became agents

AI systems now spawn sub-agents, manage dependencies, and coordinate with peers. Your compliance checklist doesn't know they exist. You need governance enforced at the data and evidence layer, not written in a PDF.

Experience

Leadership across AI governance, assurance, and regulated delivery.

2024 – Present

AI Governance Consultant

Designed and operated AI governance policy, standards, and control frameworks aligned to NSW and international expectations, including NIST AI RMF and ISO/IEC 42001 concepts. Embedded governance into procurement and vendor engagement.

2020 – 2024

Senior Manager, Digital Transformation — NSW Health (eHealth NSW)

Led AI governance, assurance, and risk management in a highly regulated NSW Health environment. Maintained AI system registers and assurance documentation.

2016 – 2020

ICT Strategy & Governance Consultant

Delivered ICT strategy and governance consulting across government, ASX-listed, and NFP organisations in Australia and APAC.

2011 – 2016

Project ICT Lead (APAC), Blackmores

Led regional technology governance and M&A support across five APAC countries, including due diligence and vendor governance.

2004 – 2011

Account Director, Telstra / NetStar

Managed compliance-sensitive government, health, and financial services accounts, leading cross-functional delivery teams.

Capabilities

Grouped by what they enable, not what they're called.

Systems & Data

SQLite PostgreSQL + RLS ChromaDB FTS5 Python CLI Tools OCR Pipelines Document Processing N8N Automation

AI & Agents

Claude Code Skills MCP Servers Agent Orchestration 7-Lens Analytical Protocol Forensic Language Audit Blind-Spot Analysis Context Engineering

Governance & Compliance

AI Governance Frameworks NIST AI RMF ISO/IEC 42001 Privacy Act (APP 10/12/13) Policy Primitives Accountability Binding Decision Trace Logging RLS Data Classification

Domain

Regulatory Compliance Evidence Management Litigation Support Veterans' Entitlements Government Technology Procurement Governance

Start with a clear conversation

If you're facing a decision where the risk of getting it wrong is high — politically, legally, operationally, or reputationally — that's the right time to talk.

After our engagement, you'll have clear accountability, explicit risk ownership, and decisions that can be defended to regulators and boards — not just documentation.

Most conversations start with:

  • “We're being pushed to move faster, but the governance isn't there yet”
  • “This sounds good on paper, but it won't survive Risk / Audit / the Board”
  • “We've already invested, and now the consequences are catching up”

I'll be direct about:

  • whether the issue is real or overstated
  • what's actually blocking progress
  • and whether I'm the right person to help

If I'm not, I'll tell you.

What happens next

  • 20-minute discovery call
  • Written assessment of fit + scope
  • Proposal with deliverables and timeline if there's a match
  • No obligation at any step

Or email me directly: mark@marmstrong.org