The Principle
I build things to run, not to babysit.
Production systems that operate without daily intervention. Observability that surfaces signal without asking for attention. Documentation that stays accurate because the workflow updates it automatically. The point of the methodology isn't the methodology. It's freedom from the systems you depend on.
That's what I think most operations leaders actually want, and almost nobody is selling it honestly.
How I Work
How a two-person firm ships enterprise work.
Twenty-five years of enterprise infrastructure. National-scale operations under regulatory frameworks like NERC-CIP and PIPEDA. Sales engineering leadership. A 36-for-36 record on competitive proof-of-concept evaluations. Top-10 finish at the AWS Enterprise Agentic AI Hackathon as the only solo competitor in a field of teams.
That's the experience. Here's the operating model.
I don't write production code by hand. I run a multi-layer build system where no single AI tool gets to be both the author and the reviewer. Architecture and planning happen in one place. Execution happens in another. Adversarial code review happens in a third. A persistent vault holds the memory of every project so no session starts from scratch. Every change is documented, every decision is reversible, every fix is grounded in an investigation written down before code is touched.
The AI gives me programmatic speed. The 25 years of enterprise infrastructure gives me the judgment to keep the speed from producing garbage. The two together are why a two-person firm can credibly take on enterprise work.
Engagements
Five ways I work with clients
Five ways I work with clients today. Each one is a different commitment level and a different shape of outcome. Most clients start small and expand if the fit is right.
Strategic AI Assessments
- For
- Executive teams making AI investment decisions in the next two quarters
- What it looks like
- A clear, unhyped read on where AI fits in your operations. What it would cost to do well. What it would cost to skip. Delivered as a written report and a working session, not a slide deck that goes in a drawer.
- Time-shape
- Four to six weeks, scoped at engagement start
POC and Evaluation Frameworks
- For
- Vendors and resellers running enterprise sales motions, plus enterprise buyers running competitive evaluations
- What it looks like
- I spent a decade designing POC programs that won competitive evaluations. There’s a version of that built with modern AI tooling that doesn’t exist in the market yet, and I can build it for you.
- Time-shape
- Six to twelve weeks, depending on scope
Observability and Telemetry Platforms
- For
- Operations leaders with messy data they can’t see clearly through off-the-shelf tools
- What it looks like
- Custom dashboards. Programmatic data ingestion and processing. Real-time signal extraction from systems that weren’t designed to be observed. Most off-the-shelf tooling stops where the interesting questions start. I build past that line.
- Time-shape
- Eight to sixteen weeks for the first production system
Production Software Builds
- For
- Organizations that need real systems built, not prototypes
- What it looks like
- AI-native where it matters, traditional where it doesn’t, compliance-aware from day one. Built to ship, run, and stay out of your way. Multi-phase delivery with documented architecture, audit trails, and a real handoff plan.
- Time-shape
- Twelve to twenty-four weeks per phase, multi-phase engagements common
Fractional Chief AI Officer
- For
- Executive teams that want senior AI leadership without making a full-time hire
- What it looks like
- Strategy. Vendor selection. Architecture review. Internal capability building. The unglamorous operational work that makes AI initiatives actually land instead of stalling. I sit alongside your leadership team, not above it.
- Time-shape
- Monthly retainer, typical engagements run six to twelve months
Proof
The deliverables aren't hypothetical.
Federal contract delivered
Multi-phase government software contract, sole developer, delivered ahead of schedule. Five of eight milestones complete in the first four months. Production systems running today.
36 for 36
A perfect record on competitive proof-of-concept evaluations during a decade in enterprise sales engineering. The methodology that produced that record is the same methodology I run now.
AWS Enterprise Agentic Hackathon
Top-10 finish in July 2025 as the only solo competitor in a field of teams. The first published artifact of the operating model that became Phi Intelligence.
Published methodology
A monthly publication on AI methodology and the operating discipline behind production AI systems. Read by senior technology leaders across North America.
Engage
Start a conversation
I'm holding capacity for a small number of engagements at a time. I'd rather have ten honest discovery calls than one premature pitch.
Vancouver, Canada