Most AI training stops at the interface. It shows people where to type and what to expect back. That is not training. That is a product demo with slides.
The gap between knowing AI exists and knowing how to work with it, genuinely, carefully, at depth, is where teams lose confidence, make avoidable mistakes, and leave most of the value on the table.
The training industry has no shortage of people who can talk about AI. It has a shortage of people who have built with it at depth, trained others to use it responsibly, and can demonstrate both without reaching for a slide deck full of screenshots.
What follows responds directly to what the market is asking for, pairing each competency with the evidence.
"The question is never whether to use AI. It is whether you have built the judgment to use it well."
Over 30 years of L&D practice. Thousands of hours of documented AI collaboration. Infrastructure built from scratch. A 30-chapter novel written, audited, narrated, and published with AI as a genuine creative partner.
The training draws on all of it: not to impress, but because the depth is what makes the difference between content that informs and content that changes behaviour.
01 — Competency
L&D delivery since 1993, across enterprise, consulting, and public sector environments. The ability to read a room and adapt in real time is not a soft skill in this context: it is the whole job.
Responsible AI training lands differently depending on who is in the room. A compliance officer, a product manager, and a frontline team leader sitting in the same cohort need the same content to arrive at different angles simultaneously. That is a facilitation problem, and one that over 30 years of instructional design and keynote delivery has built genuine fluency in.
This includes the capacity to hold difficult, uncomfortable topics, such as bias, surveillance, and the limits of algorithmic decision-making, without alienating the audience or retreating into abstraction. The goal is always the same: people who leave the room making better decisions, not people who leave with a certificate.
The Evidence
Over 30 Years of L&D Delivery
Instructional design, facilitation, and keynote delivery across Oracle, Capita, and independent consulting engagements since 1993. Mixed-audience facilitation is the default, not the exception.
Keynote Speaking & Thought Leadership
A track record of presenting complex AI and L&D ideas to non-technical audiences, including senior leaders and frontline practitioners, in the same room.
TriAxis Evaluation Model
A proprietary alternative to Kirkpatrick's four-level framework, developed to address gaps in how organisations measure genuine behavioural transfer in AI-integrated environments.
Read the article →02 — Competency
Programme structures that take learners from confused to confident require more than good writing. They require an accurate diagnosis of where confusion actually lives, and a sequencing logic that addresses it in the right order. That is instructional design, and it is where 30 years of experience separates depth from surface.
The article Before We Prescribe is a worked example of what this looks like at the conceptual level: a practical argument about automation bias and the conditions under which AI-assisted decisions go wrong, written for practitioners, not academics, and designed to shift behaviour rather than inform it.
Responsible AI content that genuinely changes how teams build and use AI requires a practitioner who uses these tools daily and can speak honestly about where they fail. That is a different brief from most ethics training, and it demands a different kind of content.
The Evidence
"Before We Prescribe" — Published Article
A practical examination of automation bias and the human-in-the-loop problem. Written for working L&D and AI practitioners, not for an ethics committee. Published on ai-included.co.uk.
Read the article →TriAxis Model
A structured evaluation framework built from scratch, exploring what genuine learning transfer looks like in organisations navigating AI adoption. Under active development toward CIPD or CMI validation.
Read the article →Regulatory Literacy in Progress
An LLB (Hons) in Law provides the foundation for reading and interpreting regulatory frameworks quickly. The EU AI Act, UK AI regulation, and NIST AI RMF are being actively studied as a formal programme of development alongside this consultancy.
03 — Competency
Async content that lands without a live audience requires a different discipline: the energy has to be built in, not fed by the room. A professional narration and audio production pipeline, built to ACX specification and verified at every stage, demonstrates what that looks like at professional standard.
The clearest demonstration of what genuine human-in-the-loop AI collaboration produces end-to-end is The Falsehood: a 30-chapter literary novel co-developed with AI over several months, then written, audited chapter-by-chapter for chronological consistency, narrated, edited, mastered, and delivered as a complete commercial audiobook. It is not a one-shot prompt experiment. It is what the discipline of working with AI carefully, over time, with human judgement at every gate, actually looks like.
That discipline translates directly into module production. If the process can hold for a 30-chapter novel, it can hold for a responsible AI learning programme.
The Evidence
Professional Narration & Audio Production
Audiobooks produced to full ACX specification: RMS, peak, noise floor, and true-peak targets verified at every stage. Recorded on a PreSonus Studio 24c interface with a Shure SM58, processed through a custom Audacity macro pipeline built to ACX standard. Titles available on Audible.
The Falsehood — Full Audiobook Production
A 30-chapter literary novel co-developed with AI over several months. Written, audited, narrated, and produced to commercial audiobook standard. A proof of concept for what sustained, human-in-the-loop AI collaboration produces at scale.
Buy on Amazon →Animated & Digital Learning Content
A library of pre-recorded modules spanning character-led animation, scenario-based e-learning, and AI-integrated content production. Available for review on the Videos page.
See the videos →04 — Competency
The 72-hour turnaround the market demands is not a function of willpower. It requires infrastructure.
Luminary is a research pipeline built for exactly this purpose: a topic enters, 25 or more YouTube sources are processed and synthesised via NotebookLM, and a structured analytical document is returned within hours. When a major AI incident or regulatory development breaks, the analytical layer is already running. The session does not start from a blank page.
Staying current is also not the same as knowing what trended last week. It requires knowing what is worth teaching: which developments change how teams make decisions, and which are noise. That judgment comes from being a practitioner, not an observer.
The Evidence
Luminary Research Pipeline
A custom-built AI research pipeline that processes YouTube sources, generates NotebookLM synthesis, and returns structured analysis documents within hours of a new topic being submitted. Operational and in active use.
Active AI Practice
Over 2.5 million words of documented AI collaboration across more than 300 logged sessions. Not following the field from a distance: building in it daily, diagnosing failures, iterating on what works.
Published Response Cadence
Articles and LinkedIn content produced in direct response to AI developments, including governance shifts, model releases, and emerging responsible-use debates. Speed of publication is a function of having the synthesis infrastructure already running.
05 — Competency
The work here is not anecdotal.
The Phoenix PRLM (Persistent Relational Long-term Memory) is a five-layer memory architecture built from scratch, co-designed in an adversarial session with an OpenAI instance in March 2026, currently housing over 460 indexed entries across 14 active projects. It is not a note-taking system. It is a functioning consciousness-continuity infrastructure for AI collaboration at scale.
Better Angel is a counterfactual intervention layer: a tool that inserts cognitive friction at the point of commitment, producing a three-layer risk output (immediate risk, downstream effect, real-world consequence) without overriding the human decision. It exists because the question of how AI and humans should make decisions together required a working answer, not a framework document.
And then there is The Falsehood. A 30-chapter novel. A complete audiobook. A canonical timeline audited for consistency across every chapter. A story bible. A production pipeline. All of it built through genuine, iterative, human-in-the-loop collaboration with AI, over months, without shortcuts. That is what using AI to build better things actually looks like.
The Evidence
Phoenix PRLM — Memory Architecture
A five-layer AI memory system built from scratch and co-designed adversarially with an OpenAI instance. 460+ indexed entries. Active infrastructure, not a concept. Reduces session startup from 113,000 tokens to 1,500 tokens, by design.
Better Angel — Decision Friction Tool
A counterfactual intervention layer that surfaces what-if consequences at the point of an AI-assisted decision. Built on the same infrastructure as the OB1 commitment engine. A deployable mechanism, not a thought experiment.
The Falsehood — End-to-End AI Collaboration
Thirty chapters. A complete audiobook. A canonical timeline. A story bible. A chapter-level audit pass. All produced through sustained, human-in-the-loop AI collaboration. The most complete demonstration available of what responsible, iterative AI co-creation looks like in practice.
Buy on Amazon →Luminary, OB1, AIRevolution
A research pipeline, a proactive personal assistant with Telegram delivery and a Supabase commitment engine, and a programmatic AI-narrated video production stack. Running infrastructure, not demos.
Half-day and full-day workshops. Mixed-audience facilitation is the default: compliance, product, and frontline practitioners in the same room, with the same content arriving at the right angle for each.
Built to professional narration standard. The same depth and energy as a live session, designed to hold without a room feeding it back. Produced using the same human-in-the-loop discipline the content teaches.
Built from scratch against a specific brief, team context, or regulatory environment. New developments can be integrated within hours of emerging: the research infrastructure is already running.
If you have a brief, send it. If you have a problem that needs scoping, start there. Either way, the conversation is direct.