AI-native engineering leadership and regulation-ready AI products for proptech, insurtech and healthtech, from a principal who knows the difference.
Principal-ledProptech · Insurtech · HealthtechLondon-region infrastructure
Why now
Ten years ago the question was the cloud. Now it’s AI: how to adopt it without being left behind, and without building risk nobody has measured.
Engineers use AI ad hoc or not at all, with no shared practice, no guardrails and no measure of whether it works.
The knowledge that would make AI useful sits in documents, tickets and old systems that nothing can reach safely.
A customer, an auditor or a regulator. In proptech, insurtech and healthtech, “the model decided” is not an answer you can give.
Services
Every engagement is led and delivered by a principal engineer. No juniors learning on your budget.
Build
From idea to production in weeks, with the controls built in from day one rather than added before go-live.
Lead
A fractional CTO for startups, or a Head of Engineering or Staff Engineer, embedded in your team for a few days a week. We bring in AI-native ways of working and leave them in place when we go.
Review
A fixed-scope, candid look at the systems you have and the AI you are using or planning.
Request a reviewApproach
Most AI problems are not LLM problems. A rule, a calculation or a classical model is often faster, cheaper and explainable by default. Watch how we route real work.
Measure how work flows today: lead time, quality, cost and risk.
For each decision, choose deterministic, classical ML or agentic, judged on latency, cost, risk and audit.
Tenancy, permissions and approvals are designed in, and every action lands in one audit trail.
Evaluation suites and live metrics show whether it works. We don’t quote invented percentages.
Accountability
What we build leaves a trail anyone can follow: which path a decision took, how long it took, what it cost and who signed it off.
See how we’d do it for youProducts
We also build our own AI products for proptech and insurtech, using the same methods. Coming soon.
About
“I’d rather tell you not to use an LLM than sell you one you don’t need.”
I’ve spent my career building systems where mistakes are expensive: insurance platforms selling over a million policies, graph-backed intelligence tools for the Home Office, and AI that has to explain itself, or at least show where its answers came from. My MSc thesis was on neural networks for hate-speech detection, and I’ve enjoyed working through every stage of AI’s evolution since.
AI is more than LLMs. I’ve built with NLP, classical machine learning such as logistic regression and support vector machines, image classification, and plain deterministic rules, and I use whichever the problem needs.
I founded Stack Foundry in 2025 so that organisations can get senior judgement on AI without hiring a Big Four bench. You work with me directly, supported by AI agent squads I run and review, not by a team of juniors.
Contact
Tell us where you are and what you need.