Writing

  • DigiRelevance tile: "Deploying AI? Show your working." beside a grid of twelve controls inside a management system frame, most evidenced and some in progress.

    Deploying AI? What UK organisations should have in place

    Most organisations are not building AI, they are deploying it. Regulators, insurers and courts are now asking them the same question: can you show your working? Here is what to have in place.

    Read more
  • Diagram: one photo, carrying its EXIF and GPS metadata, passes through two API keys (World Labs and FAL) to five generative models: marble-1.1, nano-banana, gpt-image-2, hunyuan-3d and elevenlabs-sfx.

    One image, five models: what an auditor would ask about image-blaster

    image-blaster turns one photo into a 3D scene using five generative models. Seven questions an ISO/IEC 42001 auditor would ask about it, and how to bring a tool like this inside your AI management system.

    Read more
  • Grid of 1,767 squares, one per security log event over 12 days: 1,306 failed logins in coral, 460 other events in grey, and a single green square for the one successful login.

    1,766 attempts to get into my website. None got in.

    Twelve days of WordPress security logs: 1,766 attempts to get in, a botnet that came looking for me by name, and why 2FA did the heavy lifting.

    Read more
  • Navy card reading The accountability gap, with a diagram contrasting a single central regulator against four separate sectoral regulators.

    The JCHR’s AI blueprint and the accountability gap

    The Joint Committee on Human Rights called for an AI Bill and a single statutory regulator. The government kept oversight distributed across sectoral watchdogs.

    Read more
  • The Semantic Validation Loop

    In an earlier piece (here) I explained a principle I work by: when mixing local and frontier AI in consulting work, the data stays local, but the abstractions can travel. Raw client information is processed by models you control. Patterns and architectures are discussed with frontier models. The boundary is enforced by tooling, not by…

    Read more
  • The Data Stays Local, But The Abstractions Can Travel

    Responsible AI and building trust with your clients

    There’s a question I hear more frequently now than at any point in my career: can we use AI on this project without our data leaving the building? It comes from finance directors, facilities heads, compliance teams, procurement leads – people whose lawyers have begun asking pointed questions about Article 28 processors and Data Processing…

    Read more
  • Building with AI

    AI can help you build faster – but there is a big gap between something that runs and something that works reliably for a client. This article is about what actually filled that gap.

    Read more
  • Let your organisations innovators “Vibe Code” with enterprise level compliance and safeguards

    We’ve all heard the term “Vibe coding” which has emerged as a rapid, intuition-led development style. While creators can now iterate quickly using AI, they often bypass the structure, frameworks, and compliance gates that have formed over time via tenured DevOps environments. Being very effective for prototypes and demonstrations, this approach often produces “fragile” software…

    Read more
  • How to learn from your mistakes – Deloitte doubles down on AI

    “That had to hurt” having to provide a partial refund to the Australian federal government after it’s independent assurance review was found to contain multiple errors, including non-existent references and citations. Despite this, or maybe even because if this, Deloitte is doubling down on AI across its services anyway. They’re not backing away from AI…

    Read more
  • Artificial, Not Assumed: The Real AI Playbook for Law Firms

    The most dangerous feature in legal AI isn’t hallucination, it’s assumption. As AI moves from buzzword to backbone, the “A” must stand for artificial, not assumed intelligence. That mindset shift is the single most important development in LegalTech right now, because it reframes AI as a disciplined tool that augments lawyers rather than an omniscient…

    Read more