We didn't start as a consultancy that added AI to its offering. We started inside regulated industries — understanding the legacy systems, governance constraints, procurement cycles, and internal politics that make real transformation hard. That's still our edge.
Marzal Labs was founded by people who spent years inside regulated industries — not advising on them from the outside. We know what legacy data infrastructure actually looks like at 3pm on a Friday when a regulator asks for a report. We know the procurement cycles that add 6 months to every technology decision. And we know the internal politics that kill transformation programmes before they deliver anything.
We built Marzal Labs because the organisations that need AI the most — insurers, financial services firms, NHS trusts, regulated enterprises — are consistently underserved by two types of provider: global consultancies that over-scope and under-deliver, and technology vendors that sell platforms without building foundations.
Our model is different. We are small by design, expert by requirement, and production-focused by conviction. We don't do decks without delivery. We don't build PoCs that stall. We don't recommend platforms we haven't built production systems on ourselves.
These aren't values we display in a slide deck. They're the lens through which we evaluate every architecture decision, every engagement scope, and every recommendation we make to a client.
Every system we build is designed to make your experts more powerful — not redundant. The goal is always the same: your best people spending more of their time on the decisions that only they can make. We design for HITL from the first line of architecture, not as a compliance afterthought.
A demonstration does not transform an organisation. A working system does. Every engagement we run ends with AI running in your environment — not a roadmap for what could be built, not a PoC that needs a phase 2, not a recommendation for a vendor to implement. We measure our success by what reaches production.
AI on unprepared data is not a competitive advantage. It is a liability. We always assess the data foundation before recommending any AI initiative — because that's where the real constraints are, and because building on solid foundations is the only path to AI that scales and satisfies regulators.
We were founded by practitioners who spent years inside regulated industries — not advising on them from the outside. We know the legacy systems, the governance constraints, the procurement cycles, and the internal politics that slow transformation down. That's not a weakness. That's the point.
Every person at Marzal Labs has operated inside regulated industries — not just consulted to them. That means we understand the real constraints, the real stakeholders, and the real difference between a demo and a deployed system.
Founder and Principal Architect of Marzal Labs. Over 20 years working inside the UK's most data-complex regulated environments — Lloyd's of London, specialty insurance, NHS, regulated financial services, and government. Led data and AI programmes across Nationwide Building Society, Talbot Underwriting, Gen Re, STARR, and the UK Covid-19 Test & Trace programme. His architecture-first conviction: AI on unprepared data is a liability, not an advantage. Holds an MSc in Information Technology.
AI Engineer with a Master's in Applied Linguistics, specialising in LLM workflows, agentic AI orchestration, and NLP pipelines. Applied deep expertise in language structure to develop prompt engineering frameworks, optimise model behaviour, and design natural-language interfaces. Proficient in Python and MCP prototyping, integrating linguistic insight to bridge human communication and AI systems.
Microsoft Fabric, Databricks, and Snowflake specialist with a background in insurance and financial services data architecture. Designed and built the medallion platform that unified 14 source systems for a mid-market enterprise in 16 weeks.
From first conversation to production — expertise-led, full accountability. What we scope, we deliver. Every engagement gets a named lead, full documentation, and knowledge transfer built in from day one.
A time-boxed, fixed-scope assessment that gives you a clear picture of where you are, where you need to be, and exactly how to get there. The right way to start any data or AI programme — before committing to build.
End-to-end delivery of a defined scope — a data platform, an AI agent, a governance framework. Milestone-based, expert-led, with knowledge transfer built in from day one. We deliver what we scope.
Ongoing expert architectural guidance and strategic advisory. A fraction of the cost of a full-time hire for organisations that need access to deep expertise without committing to a full programme.
We are always interested in speaking with exceptional data engineers, AI architects, and domain consultants who share our approach to production-grade AI in regulated industries.