Author and maintainer.
Andrew Moore writes and maintains Axiom Verity. He spent 25 years in board-reported technology, cyber and cryptography risk at systemically important UK institutions, including as Global Head of Cryptography at HSBC and Director, Technology & Resilience Risk at Royal London Group, and now works on AI governance, safety and security.
Why the lab is called Axiom Verity.
A starting proposition accepted as true without proof, used as the foundation for a logical system.
The input. The rule.An unassailable, real-world truth: not assumed, but proven; the actual state of things as they are.
The output. The proof.The two words aren't a redundancy: they describe a loop, not a synonym.
Axiom is the input: the deterministic controls, constraint solvers, and cryptographic validation gates engineered at Axiom Puzzleworks. Verity is the output: the audited, provable outcome those controls guarantee.
Sound axioms, rigorously applied, produce verity. That's not a coincidence of language: it's the operating model. Governing AI as a proof, not a promise.
Twenty-five years across regulated financial services.
- Founder & Managing Director · Axiom Puzzleworks (2025 – present)
- Director, Technology & Resilience Risk · Royal London Group (2023 – 2025)
- Global Head of Cryptography · HSBC Bank (2019 – 2021)
- Director, Information Integrity & Protection · Barclays Bank (2012 – 2019)
- Head of Operational Rigour · Barclays Bank (earlier career)
- IT Infrastructure Audit Manager · Barclays Bank (earlier career)
- IT Assurance Consultant · Ernst & Young (earlier career)
- Chartered Accountant · Hurst & Company (earlier career)
Certifications and professional standing.
BSc Computation, University of Manchester.
2:1 Honours.
Governed AI, running in production.
Axiom Puzzleworks is a live testbed for governed AI: an AI-integrated logic puzzle platform that combines generative model outputs with deterministic constraint solving, schema validation, automated evaluation, uniqueness testing, and human approval gates before anything ships.
Deterministic control of probabilistic systems
Constraint solving and schema validation gate every generated output before release: the discipline ISO/IEC 42001 and the EU AI Act describe, running in production.
Traceability and versioning
Model selection, versioning, and production telemetry are logged for every release, not reconstructed after the fact.
Human-in-the-loop by design
Approval gates sit between generation and publication as a structural part of the pipeline, not a policy document describing intent.
Public code, not just this description of it: see Projects for working demonstrations, each with governance notes connecting it back to the Library.
Axiom Verity and Axiom Puzzleworks are separate: Axiom Puzzleworks is the author’s other company, for tooling and puzzles, and the GitHub organisation that hosts the lab’s public code. See the independence statement for funding and affiliations.