Independent Technical Counsel — Switzerland
You've built
something real.
Then complexity wins.
Every team reaches the inflection point — the precise moment where capability and clarity diverge. Where what you've built is real, but what you can name, prove, govern, or explain is uncertain. That gap, at high stakes, is where I work.
Complexity doesn't
announce itself.
It compounds.
Your team is capable. You've shipped something that works, at scale, under real conditions. That's not in question.
But there are moments — specific, high-stakes moments — where your own expertise isn't sufficient to see it clearly. A critical architecture decision. A security posture that needs to hold under scrutiny. An AI system that needs governance before it becomes liability. A business model that needs design.
The cost of getting these wrong is too high to guess at. What you need isn't a large consultancy with overhead and agenda — it's a single senior voice who has been exactly where you are, who can name what you can't yet name, and who stays long enough for that clarity to compound.
"We're capable. We've built something real. But at this specific moment — this decision, this gap, this question we can't yet answer — we're not sure we can see it clearly enough to act on it. And the cost of getting this wrong is too high to guess."
I've been at
that edge.
I know what it feels like to stand at the inflection point — capable, building fast, sensing that something needs to be named, proved, or governed before it becomes irreversible. I've sensed architecture smell before I had the language for it. I've adopted AI tooling before there was a governance framework. I've structured data for reasoning before anyone called it a knowledge graph.
Switzerland-based, senior-level, independent. The person who diagnoses is the person who advises. The same partner who named the problem at engagement one is still there when the next inflection point arrives. No re-onboarding. No context reset.
Three steps.
One engagement at a time.
A focused engagement — review, assessment, workshop, or prototype. Short enough to be low-risk. Sharp enough to surface what matters. The output is a clear, named diagnosis. Not a list of concerns. Something you can act on immediately and stand behind in any room.
A concrete artifact — a report, a design, a working prototype, a governance framework, a knowledge graph — that the team can use immediately. No abstract recommendations. No 200-page decks. Something specific, defensible, and real.
The same senior partner stays. Context compounds. The architecture review becomes the security review. The prototype becomes the platform. The business model becomes the foundation for scaling decisions. No re-onboarding. No context reset. No limits.
The work I do.
Each one a distinct discipline.
Your system works — but you're sensing something that's hard to name. Performance that degrades at scale. Coupling that makes changes dangerous. A structure that made sense in month two but is now invisible tax on every decision. Architecture review surfaces what the team is too close to see, names it precisely, and maps a path forward that the team can act on without a rewrite.
You've built the product. The business model is still a hypothesis. Subscription or usage-based? What does the unit economics look like? How do you explain the commercial logic to an investor who didn't build it with you? Business modelling engagements produce a structure you can defend — in a pitch, a board meeting, or a pricing conversation — grounded in how the product actually works.
From knowledge graph design and GraphRAG deployment to AI-SDLC governance — the full stack of building AI systems that are trustworthy, traceable, and defensible. Not just making AI work, but making it something you can stand behind. Whether you're governing an existing AI pipeline or designing one from scratch, the output is a system that knows rather than guesses — and a team that governs rather than hopes.
Tools I've built
for problems I've had.
The same rigour I apply to client work, applied to tools and products I needed myself. Both in active development — available to explore now.
Yatzy
A Yatzy dice game built end-to-end using a proprietary AI-SDLC specification framework. A live demonstration of governed AI development — from first spec to deployed product, every decision documented and defensible.
yatzy.refork.dev → In developmentSafeScan
Native iOS QR & barcode scanner with persistent history, automatic content detection (URLs, Wi-Fi credentials, contacts), QR generation, and Pro-tier URL safety checks via Google Safe Browsing. No trackers. Your data stays yours.
safescan.refork.dev →
When complexity
stops winning,
everything changes.
"Capable but uncertain at a critical moment."
- Architecture flaws compound silently — into rewrites that could have been avoided
- Security exposure becomes a breach before it becomes a report
- Ungoverned AI becomes liability before it becomes a framework
- The inflection point passes — the window closes, the technical debt locks in, the competitor ships first
- Building fast, sensing risk, flying partly blind at the exact moments when clarity would have changed everything
"Decisive and accountable."
- You can name it — in language that holds up under scrutiny from a board, an investor, or a regulator
- You can prove it — security posture documented, architecture sound, AI governed, business model designed
- You can stand behind it — every critical decision has a named diagnosis, a concrete fix, and a partner who knows the full history
- You move forward — faster, with clarity, with a senior partner who compounds in value with every engagement
Describe the inflection
point you're facing.
"When complexity starts winning — I give you the clarity to act, the proof to stand behind it, and a partner who stays."