Recent content on Mario Thomas
On 2 August 2026, the EU AI Act’s rules for standalone high-risk systems were supposed to take effect. They will not. Six days before the deadline, the EU brought an amendment into force deferring those obligations to December 2027, because the standards and implementation machinery needed to make them workable were not ready. In this article, I set out what the deferral reveals about Boards…
Every AI business case counts the hours saved. Almost none counts the hours added: the reviewing, correcting, and supervising that AI outputs demand before anyone can rely on them. That oversight labour is real, regulation increasingly mandates it, and people absorb it silently on top of existing roles. BCG research published in March 2026 shows the consequence: a distinct mental fatigue attaching…
In the Redeployment Dividend I argued that AI’s real prize is releasing intellectual capital from undifferentiated work, not cutting headcount. The evidence has now caught up with the argument, and it is uncomfortable. Teams that deploy AI save the equivalent of five hours per person per week, yet most of that time drains into low-value work, and nine in ten executives report no measurable…
In The Great Remaking I established that businesses are being remade around AI. But “remake with AI” is not “put AI into everything”, and the difference between them is judgement. Asked recently how I decide which AI to use, I said that I do not start there. In this article, I argue for the three questions that come first, and that the one doing the real work is not the…
The market now treats visible AI adoption as proof a company is driving forward through innovation, and the chief executive is the one expected to show it. Being seen to adopt, not misleading the market, and choosing well are three demands held at once. In this article, I argue that AI does not rewrite the chief executive’s duties; it changes the conditions under which they are discharged.…
Every position in the AI Sovereignty Trilemma carries a cost, but only one is shown to a Board before it is paid. The visible cost is that sovereign capability is dearer, which is where most sovereignty conversations stop. The hidden cost belongs to the convenient alternative, frontier capability bought cheaply and governed elsewhere, and it stayed invisible only because the control it surrenders…
The case for AI in the finance function is no longer in question. Commitment to it now runs well ahead of readiness, but accountability does not wait for that gap to close. The CFO answers for the integrity of the accounts and the stewardship of capital every reporting cycle, ready or not, and AI is already inside the work that produces both. In this article, I argue that AI changes how each of…
The strategic value of data is no longer in question. The next frontier is data whose meaning and relationships are explicit enough for machines to reason over rather than merely retrieve. Unstructured information must be interpreted by whoever consumes it, whether that is a person reading a report or an AI system generating an answer. Structured information makes explicit the relationships…
The company secretary’s role was built to maintain the conditions under which directors can apply judgement and the company can meet its governance obligations. Both are now being remade: by AI tools inside board administration that compose the materials directors will judge, and by AI deployments inside the business that shape the compliance position the secretary must disclose. This…
A foundation model arrives carrying a value system its provider built: what it refuses, how it frames a sensitive subject, how it resolves a question with reasonable views on either side. That standard, not the organisation’s, is the one in force, and it changes with each model version without the Board’s consent. System prompts, retrieval, guardrails, and fine-tuning constrain the…
Updated AI models arrive almost daily, alongside new architectures and efficiency techniques. The instinctive reading is that this is good news for the AI budget, and that the capex commitments hyperscalers are making will turn out to be over-sized for a market becoming dramatically more efficient. That reading is the wrong way around. This article examines why architectural efficiency releases…
The UK regime now requires four safeguards for any significant decision taken solely by automated processing: information, representations, human intervention, contestability. On the page these are procedural rights. In practice they all depend on something the law does not name: whether the organisation can interrogate its own decisions well enough for the safeguards to work. For a rule-based…
The chair’s role was built for a stable world that no longer exists. The Board’s own work is being remade by AI tools that silently invite the substitution of director judgement, and the work the Board governs is being remade by operational AI deployments most directors cannot interrogate. This article works through how Cadbury, the FRC, and the IoD have set out chair responsibilities,…
For decades, tighter discipline over technology spend has rewarded the finance functions that applied it. AI capital behaves unlike anything they have measured before: it appreciates rather than depreciates through use, accumulates through reinvestment rather than paying back linearly, and surfaces value in functions other than the one that funded it. The project-ROI lens, optimised for…
Boards have always governed under incomplete information. What the four indicator types offer is not more information but a progressively higher quality of it. Lagging indicators establish what happened, leading indicators signal direction, predictive indicators model possible futures, and reasoned indicators prove what is certain. Applied in combination to a single decision, they represent…
Boards have always governed under conditions of incomplete information. What has changed is the volume and velocity of that information, and the speed at which AI systems now act upon it. Lagging indicators report on the past. Leading indicators signal what is likely to happen next. Predictive indicators model possible futures. But automated reasoning offers something different entirely: proof.…
The informational asymmetry between management and the Board has always been the central tension of governance. For AI, it is no longer manageable through existing structural checks; the distance is not merely larger than previous technology waves, it is qualitatively different. A director must be able to interrogate maturity claims, assess whether governance is operational or merely…
The part of AI value that is technological and replicable is also the part that standard progress measures capture best. Pilot counts, budget lines, and strategy documents say nothing about whether the essence of work is genuinely being remade, or whether the three compounding loops are operating. A Board that accepts those reports without probing them is not exercising oversight; it is ratifying…
Every previous technology wave rewarded fast followers. Identify what the leaders built, acquire or replicate it, close the gap. That logic fails for The Great Remaking — not because AI is different technology, but because the source of advantage is not a product that can be studied and replicated. It is operational accumulation: proprietary data shaped by AI-integrated workflows, human capability…
AI is not remaking the four dimensions of the essence of work at the same speed, through the same mechanisms, or toward the same end state. Treating them as a single strategic question is the mistake most organisations are currently making. The organisations pulling ahead understand which dimensions are moving fastest in their sector, where redesign would produce the greatest compounding…
An AI agent that can only see the public internet is no more useful to an organisation’s business than a very expensive search engine. The intelligence is not the constraint. The connectivity is. Model Context Protocol — MCP — is the infrastructure standard that connects agents to the proprietary data, systems, and processes that constitute real competitive advantage. This article explains what…
Version 1.5 · 19 March 2026 This paper proposes a lightweight, deployable architecture for solving the agent discovery problem in Model Context Protocol (MCP) ecosystems. The core proposal is a DNS-based convention — an _mcp TXT record — that points any compliant AI agent to an organisation’s MCP registry. DNS-based discovery is not new: MX, SRV, _dmarc, and WebFinger all use the same…
Over five decades, five technology revolutions each transformed organisations, but none restructured the essence of work itself. AI does — remaking how organisations think, decide, create, and deliver. The gap between bolting AI onto existing processes and redesigning how work is structured is already producing four times higher total shareholder returns for those who commit. This article defines…
In June 2024, I proposed that organisations would need to compensate workers whose expertise became embedded in corporate AI models. The rise of personal AI agents inverts that assumption entirely: individuals are already investing thousands annually in always-on agents that encode their professional judgement, domain expertise, and decision-making patterns — capability that belongs to them, not…
Consumers are voluntarily paying $3,650–9,125 annually for always-on AI agents — more than their combined entertainment subscriptions. When ChatGPT followed exactly this pipeline from consumer novelty to shadow enterprise adoption within three years, most organisations were caught unprepared. Agentic AI is now running the same cycle. This article examines the inference migration — the…
Boards apply rigorous stewardship to physical assets: regular condition assessments, clear ownership, maintenance investment, impairment testing. Data assets — which increasingly drive competitive advantage — receive none of these disciplines. The gap isn’t technical; it’s governance. Accounting standards render data invisible on the balance sheet, so Boards govern it as though it…
AI coding tools don’t close the expertise gap — they amplify it. Research shows senior developers capture twice the productivity gains of juniors, while a randomised controlled trial found experienced developers actually worked slower with AI than without, the hidden taxes of verification offsetting initial speed. This article explores the verification premium — and why Boards should ask not…
Workers with genuine AI capabilities command premiums of 28-56%; those targeting AI-exposed roles without substantive skill development face a 29% earnings penalty. The same roles, opposite outcomes, and the difference lies in the quality of capability investment, not access to tools. This article examines why this bifurcation extends to the Boardroom itself, where the IoD now positions AI…
AI’s primary value isn’t replacing people, it’s releasing the intellectual capital trapped in undifferentiated work. Yet in many Boardrooms, workforce reduction remains the default success metric for AI initiatives. This article makes the case for the redeployment dividend: redirecting freed human capacity toward outcome-impacting work, complex judgement, and innovation that AI…
As we return to our desks for 2026, the AI forces demanding attention aren’t distant possibilities but strategic choices already in motion. AI is embedding itself into enterprise applications faster than organisations can govern it, whilst simultaneously eroding the human capabilities needed to oversee it. In this article I examine five of these forces — AI’s shift from content…
In 2025 Boardrooms saw a collective shift in how they thought about AI’s role. What they spent 2023 and 2024 reacting to became a question of strategic investment in organisational infrastructure. They moved from “what can it do?” and “should we use it?” to “how do we navigate competing pressures and make this core to how we operate?” In this article, I…
Forty-two percent of companies abandoned the majority of their AI initiatives this year — not because AI failed, but because organisations applied generative AI to problems better solved by traditional machine learning or deterministic automation. This article examines the recalibration underway as sophisticated adopters discover that LLMs excel at specific tasks but prove expensive and unreliable…
America’s 19GW power shortfall by 2028 is forcing hyperscalers to build their own generation, but the strategic insight is what happens next: surplus capacity transforms AI infrastructure operators from energy consumers into grid actors. This article examines how distributed generation reshapes the relationship between technology companies and national grids, exploring whether the UK’s…
AI governance is fragmenting into incompatible systems. Europe prioritises trust through transparency. America pursues speed through scale. China maintains control through integration. For Boards, this isn’t about flexible compliance across markets - it’s about recognising that these systems are so incompatible that trying to serve all three means serving none well. This series…
Boards frequently overestimate AI maturity by focusing on tool deployments rather than genuine capabilities, mistaking isolated pilot successes for systemic organisational readiness. This article exposes the three patterns that create the illusion—tool-centric thinking, pilot success traps, and hype-driven metrics—and provides a diagnostic framework to reveal true position and enable targeted…
Minimum lovable governance marks a shift from episodic compliance scrambles to continuous, embedded oversight that people actually want to use. In this article I explain how governance can achieve necessary guardrails whilst earning adoption rather than resistance — like an arbour that guides growth without constraining it. For Boards, minimum lovable governance presents a practical path: the…
The Governance Paradox More than 80% of employees - including nearly 90% of security professionals - use unapproved AI tools in their jobs. Yet most organisations have AI governance policies. Comprehensive ones. Carefully documented. Rarely consulted. This paradox reveals something important: governance that exists on paper but gets routed around in practice isn’t governance. It’s…
World models mark AI’s shift toward true predictive power, allowing systems to simulate future scenarios and help businesses move from reacting to events to anticipating them. Drawing on emerging research, including Yann LeCun’s work on simulation-based intelligence, this article highlights the practical gains industries like aviation and finance are seeing in operational efficiency…
While organisations transfer decision-making agency to AI systems, accountability remains with humans, yet boards approve AI deployment without investing in the verification capability needed to ensure it. In this article, I demonstrate why this creates a strategic choice with measurable consequences: augmentation preserves expertise pipelines whilst achieving efficiency gains, but replacement…
McKinsey’s 2025 research shows whilst 88% of organisations use AI, only 23% have successfully scaled agentic systems — and even fewer integrate disciplines beyond generative, limiting value to linear gains rather than exponential growth. In this article, I expand the agentic AI definition from “generative AI in a loop” to compound loops that coordinate multiple AI disciplines…
Agentic AI has become this year’s poster child, dethroning generative AI as the technology everyone wants to discuss. Yet fundamental misunderstandings about what agentic systems actually do create barriers to successful adoption. This article demystifies the hype by revealing the core truth: agentic AI is generative AI in a loop, where the machine drives iteration instead of a human, making…
Deloitte’s 2025 survey shows 69% of boards discuss AI regularly yet only 33% feel equipped to oversee it, whilst MIT finds workers at over 90% of companies already use shadow AI without governance – exposing the execution gap between strategy and action. In this article, I provide sequenced, mutually reinforcing actions that transform the Complete AI Framework from guiding policy into…
Most organisations approach AI through accumulated business cases, hoping that individual project approvals will somehow cohere into transformation. They don’t. With 92% of companies planning increased AI investment yet only 1% achieving maturity, the gap between ambition and achievement reveals a fundamental misconception: business cases aren’t strategy. This four-part series applies…
Stanford’s 2025 AI Index shows 78% of organisations using AI, yet McKinsey finds only 21% have redesigned workflows to integrate it – revealing a governance paradox where widespread adoption yields minimal transformation. In this article, I show how the Complete AI Framework serves as guiding policy that transforms the Six Concerns diagnosis into systematic action, enabling Boards to…
The true AI governance challenge isn’t pilot failures – it’s that Boards’ six core concerns demand simultaneous orchestration yet receive sequential attention through project-level adoption. In this article, I show how these interconnected priorities form the proper diagnostic lens for AI governance, revealing why addressing them together as a whole rather than individually…
Boards are approving AI initiatives at record pace – 92% of companies plan increased investment – yet only 1% have achieved AI maturity: the gap reveals a fundamental misconception about AI strategy. In this article, I expose why accumulating business cases creates fragmentation rather than transformation, and why Boards must shift from project-level approvals to orchestrating systematic AI…
Following your AI amnesty programme, speed matters: employees who disclosed shadow AI usage expect enablement, not restriction - the post-amnesty window is critical. In this article, I provide a roadmap for transforming discoveries into governed capabilities that boost organisational productivity and reduce the risk of AI moving back into the shadows again.
With a 68% surge in shadow AI usage and 54% of employees saying they would use AI tools even if they were not authorised by the company, Boards face a governance challenge traditional compliance cannot solve. This article presents AI amnesty as an important first step to minimum lovable governance - transforming hidden risks into strategic assets whilst capturing employee-validated innovation.…
When investment banks dedicate significant research to power constraints and markets reward energy-backed infrastructure with substantial valuations, UK Boards operating with energy costs four times higher than competitors need frameworks for navigating this validated reality. This article examines how Goldman Sachs’ institutional analysis transforms energy sovereignty from policy concern to…
AI governance is fragmenting into incompatible systems — Europe prioritising trust through transparency, America pursuing speed through scale, China maintaining control through integration — forcing Boards to choose rather than compromise. In this article, I explore the sovereignty trilemma and present three strategic stances for navigating these landscapes without fracturing your strategy.