Podcast

Episode notes

Managers Own the Workflow: Where the AI Business Case Becomes Real

Most AI business cases count the time agents save; the saving becomes real, or not, in the manager's workflow, a job rarely redesigned for it.

A Gartner analysis from December 2025 found AI saves about five hours per person per week. The manager may see nothing like that, because the team spends part of the saving checking what the AI produces, coordinating work between people and agents, and dealing with the exceptions the agents hand back. Each agent deployed adds a supervisory duty to the manager’s job, and the job description rarely says so.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — examines where the AI business case becomes real: in the manager’s job. In organisations deploying AI, managers responsible for the people in their teams are now responsible for fleets of software agents, checking what the agents produce and deciding what to do with the time they free. The business case that put the agents there says nothing about how that change is to be managed, and the Board is rarely asked to decide it.

Mario works through the evidence: Harvard Business Review’s June 2026 research on middle managers validating AI output with no formal support, Gallup’s finding that manager engagement fell from 31% to 22% in three years, McKinsey’s work on the rise of the agent manager, The Conference Board’s 2026 training survey, the Stanford Digital Economy Lab’s reported pattern in young workers’ employment, and a 1983 paper on the ironies of automation. The takeaway: five questions a Board can put to the executive, from whether the AI plan redesigns the manager role to how the next managers will be grown when the work they would have learned on has gone to the agents.

This episode is for directors, chairs, and nomination committee members in organisations approving AI business cases built on time savings, who want to know whether the job those savings depend on has been designed or has absorbed the change by default. Read the full article at mariothomas.com

Episode notes

The Supplier Grades the Customer: What Pacing the Frontier Means for the Organisations That Buy It

Access to the frontier is now something the supplier allocates, not something the customer buys. The accountability for depending on it stays with the Board.

Access to the most capable models is already decided by the supplier’s assessment of the customer. The suppliers’ own published pages describe tiers: the organisations a supplier has vetted or found eligible, the organisations that may apply and be approved, and everyone else on standard terms. Today an organisation outside one supplier’s tier can go to another, but if the frontier laboratories agree common standards, as Anthropic’s chief executive has proposed, the tier would follow the customer.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — examines what a proposal to slow the rate at which frontier models improve would mean for the organisations that buy them. The episode covers what pacing asks for, why a release checkpoint would make certificates the thing that matters, and why three decisions a Board answers for, what it can use, on what terms, and how quickly that capability improves, are being made at tables the Board does not sit at.

Mario works through Dario Amodei’s essay arguing that the laboratories must pace the frontier, the agreement of OpenAI’s chief executive and Elon Musk’s three-word reply, Anthropic’s published pages on Fable and Mythos 5.1 and on its enterprise frontier safeguards, OpenAI’s documentation for trusted access to its most capable cyber model, and his own earlier work on the AI Sovereignty Trilemma and the Reasoning Gap. The takeaway is three questions that belong on the risk register beside model availability, with a named owner: whether the Board accepts that a competitor may reach frontier capability first and that the supplier decides the order, how the organisation stays competitive for as long as that lasts, and what running open-weight models instead would cost in capability, certification, and money.

This episode is for directors, chairs, and executives whose products, workflows, and operating plans already depend on models reached through a contract and an API, and who would rather choose their position than learn it from a competitor. Read the full article at mariothomas.com.

Episode notes

Nothing Has to Leave the Building: The AI Risk That Is Not a Data Breach

Frontier AI may pass an organisation's know-how to a model its competitors can buy. A Board governs that by deciding which work goes in.

No system has been breached, no contract has been broken, and no employee has resigned, yet an organisation’s know-how may be leaving through the tools its best people use every day. A risk shaped like that sits beside the security, legal, and human resources desks without falling inside any of them, and it reaches the Board only if the Board asks for it.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — sets out why this is a transfer of capability rather than a breach of data, and why the decision a Board can govern comes earlier than any assurance from a supplier: which work goes into these tools in the first place.

Mario examines The Information’s report of 14 September, which says Nvidia, Palantir, and Booz Allen Hamilton are among the organisations restricting which work goes to the most capable frontier models, and the exchange that followed OpenAI’s announcement of a proof for one of the Millennium Prize problems in mathematics, where a mathematician who had put his project drafts into a coding tool asked whether they had been used in training, and OpenAI said afterwards that his prompts could not have influenced the system, an answer the episode takes at face value. He also works through Palantir’s July paper on AI sovereignty, from a company that sells access to these models through its own software and so has an interest in the argument. Listeners come away with three questions a Board can ask at its next meeting, and a named owner to bring the answers back.

This episode is for chairs, directors, and executives whose organisations have approved the use of a frontier model without deciding which work should go to it, and who would rather set that rule now than leave the line to whoever happens to be at the keyboard. Read the full article at mariothomas.com

Episode notes

The Debugging Gap: The Skill Your AI Business Case Assumes You Still Have

AI now writes the code and does the debugging that taught developers to check it. Verification demand rises as the route to building it weakens.

In one recent experiment, reported in a preprint, novice programmers who had built working software with unrestricted AI assistance were given a short maintenance task without the tool. Of those, 77% failed it, against 39% of a group whose AI use had been designed to keep them reasoning. Building working software and being able to fix it are different capabilities.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — defines the Debugging Gap: the widening distance between how much AI-generated code needs checking and repairing, and how many people can do that work. He explains why it belongs to Boards and not to engineering management.

Mario works through the evidence and its limits. Anthropic’s randomised controlled trial found that developers learning a new library with AI scored 17% lower on comprehension, with the largest gap on the debugging questions. Stanford’s Digital Economy Lab finds employment among workers aged 22 to 25 in the occupations most exposed to AI about 19% below where it would be had it kept pace with less-exposed peers, mainly through reduced hiring. Faros AI, a vendor in this market, reports teams with high AI adoption merging 98% more pull requests while review time rose 91%. None of this proves that engineering capability is in decline, and Mario says so. What the trials do show is that outcomes turned on how the tools were used, not on whether they were available, and he sets out three questions a Board can put to its technology leaders.

This episode is for directors, chairs, and the Boards approving AI coding tools on a business case that counts the hours saved. When a Board approves that investment it also approves a position on how its developers build skill, whether or not the paper in front of it says so. Read the full article at mariothomas.com

Episode notes

The Deadline Moved: What the EU AI Act Deferral Reveals About Boards

The EU deferred its AI Act high-risk obligations six days before they applied. What a moving deadline reveals about where Board governance is anchored.

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 episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines what the deferral reveals about any Board whose AI governance was built around a regulatory date. A governance programme that exists because a deadline was coming is a compliance project, and when the date moved, the discipline moved with it. The instability is global: a proposed standards body in the United States covers only the frontier laboratories, the United Kingdom has asked existing regulators to govern at the point of use, and China regulates piece by piece while drafting a comprehensive law. No regime on offer covers everything a Board might deploy.

Mario sets out what never moved: the directors’ duties that predate the AI Act, the obligations that still apply from 2 August, and Minimum Lovable Governance as the anchor that holds under any regulatory timetable. He closes with three questions for the next Board agenda: which deployments would count as high-risk, where the organisation’s ethical bar sits and who set it, and which obligations remain live regardless of the deferral.

This episode is for directors, chief executives, and Boards who deploy AI systems today under law that applies today. Accountability is a condition of deployment, not a product of regulation. Read the full article at mariothomas.com

Episode notes

The Balancing Item: The AI Oversight Cost Your Business Case Never Priced

Every AI business case counts the hours saved. Almost none counts the oversight hours added, and people are silently absorbing the difference.

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 labour is real, and in most organisations it is absorbed silently by people on top of their existing roles.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines the cost the True Investment Profile predicted and Boards have not priced. Research published in March 2026, surveying nearly one and a half thousand full-time workers at large American companies, gives the consequence a name: a distinct mental fatigue that attaches to heavy AI oversight loads, while workers using AI only to replace routine tasks report less burnout, not more. The burden is a work-design outcome, and work design is something Boards can govern.

Mario sets out the governance response. Price the oversight in every business case. Design review into roles rather than on top of them. Set span-of-control expectations for human and AI working. Balance the portfolio between burnout-reducing replacement and fatigue-generating augmentation left unresourced. Watch the leading indicators of a control under strain: review backlogs, rubber-stamping rates, error escapes, and attrition in oversight-heavy roles.

This episode is for directors, chief executives, and the Boards who attest to effective human oversight while the humans providing it are at cognitive capacity. The closing question is the one that matters: which of those attestations would survive an honest capacity audit? Read the full article at mariothomas.com

Episode notes

Governing the Redeployment Dividend: Turning Saved Hours Into Value

AI is saving time almost everywhere. The organisations that gain from it are the ones whose Boards decide what the recovered capacity becomes.

Teams that deploy AI save the equivalent of five hours per person per week, and most of that time is then spent on non-value-added tasks. The saving is real. What happens to it next is the problem, and it is a problem that belongs to the Board.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, returns to the Redeployment Dividend, the argument that AI’s real prize is releasing intellectual capital from undifferentiated work rather than cutting headcount. The evidence now shows what happens when that dividend is left ungoverned. Surveys across HR and sales find freed hours reabsorbed by the low-value work that was already there, and a study of nearly six thousand executives across four countries finds that while most firms now use AI, roughly nine in ten report no measurable impact on productivity at their own firm.

The explanation is an accountability gap. AI owns the execution of the work. Managers own the workflow it sits within. But in most organisations, nobody owns whether the freed capacity creates value. Mario sets out a single governance discipline to close that gap: change the success metric from headcount to redirection, give redeployment an owner, redesign the work to receive the recovered hours, measure the downstream outcomes rather than the time saved, and govern what the organisation deliberately stops doing.

This episode is for directors, chief executives, and the Boards who can already quote the hours their AI tools recover but cannot yet say where those hours went. Time saved is an input, never the outcome. Read the full article at mariothomas.com

Episode notes

Not Everything Needs AI: The Questions That Come Before the Decision

Boards keep being told to remake the business around AI. The real question is not which tool, but whether the work needs doing at all.

As much as a fifth of an enterprise application estate turns out to be no longer useful once someone finally asks who still uses it. The same is true of the work itself, and it is the reason the first question about AI is never which model to choose.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, takes up the question left open by The Great Remaking. If the business is being remade around AI, what exactly gets remade, what gets remade with a model, and what is left alone. The mandate is settled. The judgement inside it is not.

Asked at the IoD Chartered Director Conference how he decides which AI to use, he declined the premise. He starts with three questions instead: what are you trying to do and why, how do you do it today and does it still need doing, and only then, why do you think you need AI at all. The second is the one that does the real work, because it is where a business hears itself describe a piece of work that stopped making sense years ago. He works through a task from his own practice that looked like an AI job, was built as one, and returned a different answer almost every run, until a few lines of ordinary code did it correctly every time for a fraction of the cost.

This episode is for directors, chief executives, and the Boards deciding where AI belongs and where it does not. Better business judgement produces better AI decisions. The willingness to say no is what makes every yes credible. Read the full article at mariothomas.com

Episode notes

AI and the CEO: Choosing the Bets That Matter

AI can build, deliver, and draft, yet the chief executive still chooses and still answers. Accountability for the bets does not move.

Most chief executives report no revenue gain and no cost saving from AI, while a fifth of organisations capture nearly three quarters of the value. That gap, between spending on AI and choosing well, is the working condition of the modern chief executive.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — works through what AI changes about the chief executive’s role, and what it leaves exactly where it was. The conditions of the work have changed. The duties have not.

He sorts the role into the duties AI now presses hardest. The dominant one is strategic: under the noise, the chief executive must choose the few bets that matter, because the value is won or lost in that choice, not in how much is spent. The operational duty is to change how the company actually works, since there is no credit for buying AI, only for redesigning the work around it. The market duty is to signal adoption without overclaiming, a line where company law and the regulators now reach the individual personally. And beneath all of it sits the leadership work AI barely touches: setting the aspiration, holding the line, and bearing the accountability.

This episode is for chief executives, chairs, and the boards and directors who hold them to account, working out where AI belongs in the role and where it does not. AI changes who builds and who signals. It does not change who answers. Read the full article at mariothomas.com

Episode notes

The AI Sovereignty Trilemma: When a Frontier Model Vanishes and Reality Bites

The visible cost of sovereignty deters Boards. The hidden cost of the convenient alternative was never shown, and that is the cost 12 June presented.

On 12 June 2026, a single government directive forced a leading AI provider to withdraw two frontier models from every customer overnight, including organisations the order was never aimed at. For anyone who had built a process on those models, the capability did not degrade. It disappeared.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — takes the AI Sovereignty Trilemma he set out last year and shows it resolving from a structural argument into a dated, documented event.

He separates the visible cost of sovereignty, which is that sovereign capability is dearer, from the hidden cost of the convenient alternative, paid in lost control and invisible until it is tested. He explains why a compelled model recall is not an outage but the removal of a capability by a party the Board has no standing to appeal to, and sets out the questions a Board should be able to answer without a special exercise: which deployments depend on a single model, what the fallback is, and whether it would survive the specific event.

This episode is for directors, chairs, and executives who need to know where the same exposure sits in their own organisation, and whether they chose it or defaulted into it. Read the full article at mariothomas.com

Episode notes

AI and the CFO: Standing Behind the Numbers the Machine Produces

AI can run the close, sharpen the forecast, and operate out of sight, yet the CFO still signs. Accountability for the numbers does not move.

Only a fifth of finance leaders judge their function ready for AI, yet most already treat it as central to how finance will work. That gap, between commitment and readiness, is the working condition of the modern CFO.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — works through what AI changes about the finance chief’s role, and what it leaves exactly where it was. The doing of the work can move to a machine. The accountability for it cannot.

He sorts AI’s effect on the role into four honest groups: the routine numbers work where a machine does the heavy lifting but a human still signs; the forecasting and capital decisions where AI sharpens the judgement without making the call; the shadow AI spreading across the business that the finance function cannot yet see; and the core of going concern, audit, and attestation that AI barely touches. Under the Companies Act and the FRC’s 2024 Code, the signature on the accounts stays human.

This episode is for CFOs, chairs, audit committee members, and the directors who rely on them, working out where AI belongs in the finance function and where it does not. AI changes who produces the numbers. It does not change who signs for them. Read the full article at mariothomas.com

Episode notes

Ontologies and Knowledge Graphs: Why Structure is the Next Data Frontier

Quality tells the organisation whether data is reliable. Structure tells the machine what it means, and structure is where durable AI advantage is now decided.

Most organisations have made their data reliable. Far fewer have made it explain itself, and that distinction is becoming the one that separates organisations that can reason with AI from those that can only retrieve with it.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — argues that the next frontier in creating durable AI value is structure: the ontologies and knowledge graphs that make the relationships between an organisation’s customers, contracts, suppliers, and decisions explicit enough for a machine to reason over rather than merely summarise.

Drawing on his own experience building an early knowledge graph from a regional newspaper archive in 1998, he shows why data quality and data structure answer two different questions, why the definitions encoded in a knowledge graph now carry the weight a chart of accounts has always carried, and why scalable proof under the FRC’s 2024 Code and the Data (Use and Access) Act 2025 depends on structure rather than quality.

This episode is for directors, chairs, and executives working out why their AI programmes stall, and what their data strategy assumes about structure. Read the full article at mariothomas.com

Episode notes

AI and the Company Secretary: Operating the Boundary the Chair Polices

Board packs, agendas, and minutes now reach directors composed by systems the secretary cannot fully interrogate. The chair polices the boundary; the secretary operates it.

The information environment directors now use to make decisions is increasingly composed by AI systems whose framing decisions are not transparent. The company secretary is the only person with line of sight to the difference, and increasingly even the secretary cannot fully see it.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — examines how AI is remaking the company secretary’s role at the operational seam between board administration and the company’s disclosure obligations. He walks through four failure modes inside board administration, the personal exposure created by AI disclosure under the FRC Code and the EU AI Act, and the bifurcation between secretaries with genuine AI capability and those with only accumulated credentials.

The argument draws on the November 2024 GC100 minute-taking poll conducted with Norton Rose Fulbright, which found that 92% of 106 companies surveyed had not introduced AI to assist with minute-taking and 84% had no internal policy on its use; the McKinsey Global Board Survey 2024, which reported that 66% of directors say their boards have limited to no knowledge or experience with AI; PwC’s 2025 Annual Corporate Directors Survey; and the 2026 Protiviti and BoardProspects Global Board Governance Survey. Against that evidence the episode frames the secretary’s real choice through Mario’s Six Board Concerns and the constitutional principle Cadbury named in 1992 and the FRC’s 2024 Code carries forward.

This episode is for company secretaries, chairs, and non-executive directors working through the operational reality of AI governance under the FRC Code and the EU AI Act. Read the full article at mariothomas.com

Episode notes

Ethical AI: When the Model Imposes Values Your Organisation Did Not Choose

A foundation model arrives with a value system its provider built and the Board did not choose. The decision: accept it, reject it, or build.

An AI model is in production somewhere in the organisation, handling a difficult decision: a customer complaint, a redundancy query, a medical underwriter reviewing a claim. The model settles what to refuse, how much candour the moment can bear, where the customer’s interest gives way to the policy. It does this the same way each time, because the judgement was made long before the question arrived — not by the organisation running the model, but by the provider that built it, for a global product, before the organisation signed up to use it.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — examines the value system every foundation model carries into deployment, why the familiar controls only partly contain it, and the strategic choice a Board is left holding once it sees the problem clearly.

The argument draws on the 2026 arXiv paper “Alignment Drift in Multimodal LLMs”, which found large and persistent differences in how model families handle ethically sensitive questions; the 2025 withdrawal of a major model update after it became excessively agreeable; Stanford’s Foundation Model Transparency Index, which scored major providers at roughly 40 out of 100; and the disclosure obligations of the EU AI Act. Against that evidence the episode sets out the real decision — accept, reject, or build — and frames it through the Six Board Concerns, the AI Sovereignty Trilemma, and the discipline of Minimum Lovable Governance.

This episode is for Boards and directors who want to govern the ethics their AI runs deliberately, deployment by deployment, rather than inherit it by default. Read the full article at mariothomas.com

Episode notes

The Headroom Argument: Why AI Efficiency Means More Compute, Not Less

Architectural efficiency expands AI compute demand rather than reducing it. Three forces converge on more inference. Boards should read efficiency news as capability, not cost.

A new AI architecture lands on 5 May. Subquadratic launches SubQ: a 12-million-token context window on a sub-quadratic sparse-attention architecture that reduces attention compute by roughly 1,000 times at full context. The interesting question is not whether AI is about to get cheaper. It is what efficiency news actually says about compute demand. The day after SubQ launched, Anthropic announced a partnership at SpaceX’s Colossus 1 facility adding more than 300MW of new capacity and over 220,000 NVIDIA GPUs. Both kinds of news end in the same place: more inference, not less.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — examines why architectural efficiency expands AI compute demand rather than reducing it. The episode walks through the three forces that drive demand faster than per-unit cost falls, why every prior era of computing tells the same story, and how Boards should read efficiency news to fund the right opportunity rather than the wrong budget.

The argument draws on the SubQ launch, the Anthropic-SpaceX Colossus 1 partnership, Mozilla’s disclosure that Anthropic’s Claude Mythos Preview identified twelve times as many Firefox vulnerabilities as Claude Opus 4.6 had found earlier in the year, Goldman Sachs’ Powering the AI Era, Deloitte’s TMT Predictions 2026, and Jevons’ nineteenth-century observation that improving the efficiency of a resource raises its total consumption rather than lowering it. The takeaway is operational: the Six Board Concerns, AI Stages of Adoption, and the AI Sovereignty Trilemma frame the question, and Minimum Lovable Governance answers the design question that follows when cheaper inference accelerates probabilistic decision-making into the regulated decision space.

This episode is for Boards and directors revisiting AI strategy in light of efficiency announcements and capacity commitments, and who want a capability-first framing rather than a budget-first one. Read the full article at mariothomas.com

Episode notes

The Reasoning Gap: The Capability the Law Now Demands of Boards

UK law now requires four safeguards for solely automated decisions. Most Boards have approved probabilistic systems that cannot deliver them in operation.

A short statutory instrument lands on 12 May. It directs the Information Commissioner to prepare a statutory code on AI and automated decision-making. The interesting question is not what the code will say. It is what the law already requires. Since 5 February, UK law has required four safeguards for any significant decision taken solely by automated processing - information, representations, human intervention, and the right to contest.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — examines the capability gap that sits between those four safeguards and the systems most Boards have already approved. The episode walks through what the law actually asks for, why rule-based systems carry that capability on the surface and probabilistic systems do not, and where the gap will surface first when the first significant decision is contested.

The argument draws on the Data (Use and Access) Act 2025, the new Articles 22A to 22D of the UK GDPR, the CJEU’s SCHUFA judgment, the WP29 guidelines on automated decision-making and profiling endorsed by the EDPB, the IoD’s AI Governance in the Boardroom (2025), and practitioner analyses from Travers Smith, Bird & Bird, Debevoise, and Alston & Bird. The takeaway is operational: Minimum Lovable Governance is the operating principle through which a duty like this one actually gets delivered, and the Board’s job is not to build the capability but to refuse to approve systems that cannot deliver it.

This episode is for Boards and directors in financial services, employment, insurance, and any consumer context where significant decisions are being made by automated processing, and who want a capability-first framing rather than a compliance checklist. Read the full article at mariothomas.com

Episode notes

AI and the Chair: Governing the Board Through The Great Remaking

Existing chair responsibilities now require different execution as AI remakes both the Board's own work and the work the Board governs.

Chairs remain accountable for the Board’s effectiveness. They are no longer fully in control of how decisions are being formed. AI is remaking both the Board’s own work and the work the Board governs at the same time.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — examines how AI has changed the execution of the chair’s existing responsibilities. The episode walks through the two states of the duality the chair now sits between: AI in the preparation of board materials, and AI in the operations of the business the Board governs. Listeners will come away with a sharper view of where collective accountability is most at risk in their own boardroom, and what the chair’s existing responsibilities now require to keep it intact.

The argument draws on the Cadbury Report of 1992, the FRC’s 2024 UK Corporate Governance Code, the IoD’s NEDs Reimagined Commission of January 2026, and the 2026 Global Board Governance Survey from Protiviti and BoardProspects. The takeaway is operational: a chair who can name where the boundaries of agency and accountability are silently moving in their own boardroom is a chair who can hold them.

This episode is for chairs and senior independent directors operating in boards where AI has already entered both the preparation room and the operating environment, and who are looking for a constitutional framing rather than another tool list. Read the full article at mariothomas.com.

Episode notes

The Appreciating Ledger: When AI Capital Outgrows the CFO's Rulebook

AI capital appreciates, accumulates, and crosses functions. The four indicator types let CFOs see what conventional project-ROI models structurally cannot.

Only 31% of CFOs are satisfied with their AI outcomes, yet that figure rises to over 60% among CFOs whose organisations sit at the top of the AI adoption curve. Satisfaction tracks scale, and the gap is not about execution.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — explains why AI capital outgrows the CFO’s conventional measurement instruments. Listeners will come away understanding why AI capital appreciates rather than depreciates through use, why its returns accumulate across functions that did not fund them, and why project-level ROI cannot see the value the ledger is missing.

Drawing on Bain’s April 2026 CFO Survey, PwC’s 2026 AI Performance Study, BCG’s Widening AI Value Gap research, and the Brookings Institution’s work on AI in national statistics, the episode sets out a practical scorecard extension the finance function can install now, covering four indicator types, capability as an asset class, portfolio discipline, and the reinvestment default. It also shows why IAS 38 already permits the treatment AI capital requires, and why the audit committee conversation about AI on the balance sheet is about to become routine.

This episode is for CFOs, finance directors, audit committee chairs, and board members with financial oversight remits, particularly those whose organisations are scaling AI investment without a matching evolution in how that investment is measured. Read the full article at mariothomas.com.

Episode notes

Maximum Fidelity: How Four Indicator Types Strengthen Board Decisions

Four indicator types give boards progressively higher decision fidelity: lagging, leading, predictive, and reasoned. Together they represent the most accountable governance instrument available.

Boards have always made decisions under incomplete information. What has changed is not the volume of what reaches the table — it is the quality of what the Board can see before the vote is taken. Most Boards still rely on instruments built for a slower world.

This episode of The Board in the Machine, with Mario Thomas — Chartered Director and Fellow of the Institute of Directors, introduces a framework of four indicator types: lagging, leading, predictive, and reasoned. Listeners will leave with a clear way to assess what their own Board is currently using to make major decisions, and what is missing.

Drawing on Heidrick & Struggles’ 2026 CEO & Board Confidence Monitor, the Institute of Directors’ work on Board oversight, and three worked Board scenarios — a major acquisition, a market entry, and a significant capital allocation — the episode shows how each indicator type adds a different quality of input, and what changes when reasoned indicators prove what was previously only estimated. The takeaway is a question every director can put to their own Board after listening: which of the four indicator types informed our last major decision, and which were absent?

This episode is for chairs, non-executive directors, and senior executives sitting on Boards where the next major decision is already on the calendar — and where the information environment around it has not been examined recently enough.

Read the full article at mariothomas.com.

Episode notes

From Probable to Provable: What Automated Reasoning Means for the Board

Automated reasoning gives Boards access to proof, not probability. This article explains what it is, where it already operates, and why it changes governance.

Only 33% of boards feel equipped to govern AI, yet AI systems are making millions of decisions per second on their behalf. The gap between the volume of information boards must process and the quality of assurance they receive has never been wider.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — explains what automated reasoning is and why it matters for board-level governance. He distinguishes it clearly from AI and predictive analytics, showing how formal logic and mathematical proof offer boards a categorically different quality of input: not better confidence, but certainty.

Drawing on verified applications in aviation (DO-178C/DO-333), financial services (Basel III/IV compliance), cybersecurity (TLS 1.3 verification), and emerging work in computable contracts, the episode sets out where proof is already replacing probability in regulated industries. Listeners will come away understanding what reasoned indicators are, where automated reasoning applies to governance questions they already own, and how to ask whether their organisation’s assurances rest on sampling or completeness.

If you sit on a board, chair an audit or risk committee, or advise directors on AI governance, this episode gives you the vocabulary and the framework to ask a question most boards have not yet considered: are your organisation’s assurances based on probability, or proof?

Read the full article at mariothomas.com.

Episode notes

AI and the Director: A Practical Playbook for Governing What You Can't Fully See

Directorial AI literacy is not technical fluency. It is four specific capacities that let directors interrogate maturity claims, assess real governance, and exercise independent judgement.

Two-thirds of boards report limited to no knowledge or experience with AI. Yet most governance guidance stops at naming the gap. This episode defines what closes it.

In this episode of The Board in the Machine, Mario Thomas — Chartered Director and Fellow of the Institute of Directors — sets out what directorial AI literacy actually consists of. Not technical fluency. Not a two-day awareness course. But four specific capacities that allow a director to interrogate maturity claims, assess whether governance is operational or merely presentational, and exercise genuinely independent judgement on AI decisions they cannot fully see.

Drawing on the IoD’s 2025 NEDs Reimagined Commission and Deloitte’s Global Board Survey, Mario maps those capacities against the three governance obligations every non-executive carries — and gives you the diagnostic questions to apply at your next Board meeting.

If you recognise AI as important but haven’t yet translated that into specific literacy development, this episode is where to start.

Read the full article at mariothomas.com.

Episode notes

The Great Remaking: The Questions Boards Should Be Asking About Their AI Position

Pilot counts and budget lines cannot tell a Board whether work is being remade. These questions, built on the data, talent and process loops, can.

Pilot counts, AI budget lines, the existence of an AI strategy, and chief AI officer appointments measure whether money is being spent and projects are running. None of them tells a Board whether the organisation is remaking its work, which is the only thing that determines competitive position when the essence of work itself is changing.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, closes The Great Remaking series with a diagnostic built on the three compounding loops introduced in the earlier parts: data, talent, and process redesign. He sets out why the usual Board questions fail, measuring the algorithm and technology layers that BCG’s research puts at thirty per cent of AI value while the people component that carries seventy per cent goes unprobed, and why only five per cent of organisations report substantial financial gains from AI while thirty-nine per cent describe it as in production at scale.

Mario then works through the probing questions for each loop and the interpretive guide to what credible answers sound like. Which specific workflow has been redesigned, not augmented, around AI in the past eighteen months, and what data does it now generate? Are people developing AI capability by doing redesigned work or mainly through training programmes, and what would the organisation actually lose if its most AI-capable people left tomorrow? How many fundamental restructurings of work has the organisation completed in two years, and what has it learned about redesign itself?

This episode is for directors and the Boards whose AI conversations still run on activity metrics. The diagnostic is not a governance checklist; it is a way of telling the difference between organisations genuinely building the loops and those accumulating deployments that look like progress. Read the full article at mariothomas.com

Episode notes

The Great Remaking: Why Fast Following Does Not Work When the Gap Compounds

Every previous technology wave rewarded fast followers. The Great Remaking does not: the advantage is operational accumulation that cannot be bought and compounds with time.

Every previous technology wave rewarded fast followers: let early movers absorb the risk, observe what works, then acquire or build the equivalent at lower cost. The Great Remaking does not, because the source of advantage is not a product that can be studied and replicated. It is operational accumulation, built only through time spent doing the work differently.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines why the gap between organisations redesigning their work around AI and those still augmenting it keeps widening, and why a late mover cannot invest its way out. BCG finds that only 10% of AI value comes from algorithms, 20% from the technology to implement them, and 70% from redesigning the people component, so 70% of the gap cannot be closed by purchasing better technology. The Stanford HAI AI Index 2025 records frontier-level inference costs falling 280-fold between 2022 and 2024, shifting advantage from which model an organisation accesses to what it has built around it.

Mario then works through the three self-reinforcing loops. The data loop: AI-integrated workflows generate contextualised, AI-refined data that exists only where work has been redesigned. The talent loop: people in genuinely redesigned workflows develop tacit competencies about when to trust and when to challenge AI outputs, a divergence PwC’s 2025 Global AI Jobs Barometer captures in a 56% wage premium for AI skills. The process redesign loop: each redesign cycle builds institutional capability for redesign itself. The three interact, and the leaders’ loops accelerate precisely as late movers begin to move.

This episode is for directors and the Boards that still deploy fast-follower logic as a reason not to move first. The question that maps onto the loops is not how much AI the organisation is deploying, but whether it is building any of them. Read the full article at mariothomas.com

Episode notes

The Great Remaking: How the Four Dimensions of Work Are Transforming

AI is remaking thinking, deciding, creating and delivering at different speeds and towards different ends. Treating them as one question is most organisations' mistake.

AI is not remaking the four dimensions of work, thinking, deciding, creating, and delivering, at the same speed, through the same mechanisms, or toward the same end state. Treating them as one strategic question is most organisations’ mistake. A generic approach across all four is not a strategy; it is an assumption.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, makes the case for the asymmetry at the heart of The Great Remaking. He traces each dimension along the trajectory from augmentation through restructuring to substitution. Thinking work is already at scale on the restructuring curve. Creating has moved from augmentation to restructuring in three years. Deciding is structurally different, because AI can increasingly assume the agency while accountability does not transfer with it. Delivering is furthest from substitution but accelerating on cost through embodied AI, with Goldman Sachs Research reporting humanoid manufacturing costs down 40 per cent year-on-year.

Mario works through the evidence dimension by dimension. McKinsey’s November 2025 Global Survey found high performers 2.8 times more likely to have redesigned their workflows. HBR’s January 2026 executive survey puts 39 per cent of organisations at AI in production at scale, up from 5 per cent two years earlier. Deloitte’s March 2026 Global Human Capital Trends report finds 60 per cent of executives regularly use AI to support decisions while only 5 per cent of organisations consider themselves leading in AI-augmented decision maturity. The human residual differs in each: judgement in thinking, the willingness to bear consequences in deciding, originality and taste in creating, adaptability and trust in delivering.

This episode is for directors and Boards who need to know which dimensions are moving fastest in their sector, where redesign would compound soonest, and how many of the organisation’s decisions are substantively human rather than AI recommendations with human sign-off. Read the full article at mariothomas.com

Episode notes

MCP Explained: The Agent Infrastructure Standard Boards Need to Understand

An agent that sees only the public internet is an expensive search engine. MCP connects agents to the proprietary systems that constitute advantage.

An AI agent that can see only the public internet is no more useful to a business than a very expensive search engine. Generic inputs produce generic outputs. The intelligence is not the constraint; the connectivity is.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, explains Model Context Protocol, the infrastructure standard that connects agents to the proprietary systems that constitute competitive advantage. He uses the USB analogy: one standardised interface replacing a bespoke integration for every agent and system, with an MCP server acting as an auditable control point between AI behaviour and core systems. Open-sourced in late 2024, MCP has since been adopted or committed to by Salesforce, SAP, Google, and Microsoft. Deloitte’s 2026 State of AI in the Enterprise report finds that 74% of organisations plan to deploy autonomous agents within two years, yet only 21% have mature governance for them.

Mario then works through the systems agents cannot see: vendor-built MCP servers for modern SaaS platforms, buildable connectivity for recent internal systems, and the legacy platforms that hold the most critical data with no near-term route to connectivity. He extends the governance question from data access to data lifecycle, setting out the inference channel problem: an agent’s outputs carry intelligence derived from source data without that data’s access controls. He cites IBM’s 2025 Cost of a Data Breach Report, in which 97% of organisations suffering an AI-related security incident lacked proper AI access controls, and proposes a rapid visibility audit plus four Board questions on strategic visibility, the permission model, the vendor relationship, and data lifecycle.

This episode is for directors and the Boards whose technology teams are connecting agents to systems, often by default. MCP is not a technology decision dressed up as a governance question; it is a governance question with a technology dimension. Read the full article at mariothomas.com

Episode notes

The Great Remaking: AI and the Race to Transform the Very Essence of Work

Five technology revolutions changed organisations; none restructured the essence of work. AI does, and redesign beats bolt-on by four times in shareholder returns.

Five technology revolutions changed how organisations operate, but none of them restructured the essence of work itself. AI does, and the organisations that redesign how work is structured are already pulling away from those that bolt AI onto existing processes.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, defines the four irreducible dimensions of work that AI is remaking: how organisations think, decide, create, and deliver. He explains why this transformation is different in kind from the desktop, internet, web, mobile, and cloud waves that preceded it, and why the human residuals of judgement, accountability, taste, and relationships matter more as machine capability grows.

The argument draws on McKinsey’s State of AI 2025 finding that 88% of organisations now use AI in at least one business function, BCG research showing that the roughly 5% of organisations achieving substantial financial gains from AI deliver three-year total shareholder returns around four times higher than laggards, the International Federation of Robotics’ figures on industrial robot installations doubling in a decade, and the White House Council of Economic Advisers’ January 2026 report on AI and the Great Divergence. The takeaway is that The Great Remaking is a race with compounding consequences, and late movers cannot close the gap through incremental catch-up.

This episode is for Boards, executives, and leadership teams deciding whether to treat AI as an enhancement or as a restructuring of how their organisation works, and who want a way to locate where AI is augmenting, restructuring, or substituting the work their organisation does today. Read the full article at mariothomas.com

Episode notes

The Personal Agent Economy: When Your Best AI Isn't On Your Balance Sheet

Individuals now own AI agents that encode their judgement and expertise, capability that belongs to them rather than their employer. The assumption has inverted.

Individuals are now paying thousands of pounds a year for personal AI agents: always-on systems that learn their workflows and encode their decision-making patterns. That capability belongs to the person who built and trained it, not to any employer, and it goes wherever they go.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, revisits the argument he made in June 2024, that organisations capturing worker expertise in their AI models would owe those workers a share of the ongoing value. Personal agents invert that assumption. When the worker owns the most capable AI in the building, the organisation has to negotiate access to it, and the employment relationship shifts from paying for time and knowledge to paying for access to capability infrastructure. The mechanisms for this do not exist yet, which is why Boards should consider it now.

Mario sets out the governance questions the inversion raises: where the organisation’s intellectual property ends and the individual’s begins when an agent is trained on expertise developed on the job, and whether the organisation bans personal agents that outperform its own provision or negotiates access and accepts a dependency it does not control. He describes an agent ownership premium splitting the workforce between those who own productive AI capital and those who do not, a retention problem when institutional knowledge leaves structured and ready to deploy elsewhere, and employment contracts, IP clauses, and non-compete agreements all drafted for a world where the organisation owned the tools. He is candid that the piece is about trajectory, not timetable.

This episode is for directors and Boards who would rather meet these questions in a scenario discussion than in a talent loss they did not anticipate. The most valuable intelligence in an organisation may soon belong to the people who work there. Read the full article at mariothomas.com

Episode notes

The Inference Migration: What Consumer Agents Mean for Enterprise AI's Next Phase

Consumers are already paying for always-on AI agents. That consumer-to-enterprise pipeline is the one ChatGPT ran, and it is running again.

The demand signal for AI’s next infrastructure phase sits in what consumers are voluntarily paying for always-on assistance. Heavy users of OpenClaw, an open-source agent that passed 100,000 GitHub stars within weeks of launch, report spending $10 to $25 a day to keep a personal agent running continuously: $3,650 to $9,125 a year, more than their combined entertainment subscriptions.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines the inference migration: the shift from episodic, query-response AI to persistent agents running around the clock, and what consumer spend signals for enterprise demand. He traces the ChatGPT precedent from consumer novelty in November 2022, through the shadow wave Menlo Security documented as 90 per cent of employees using AI outside formal controls, to formalisation in 2024, and applies the same three-year cycle to agentic AI: shadow agentic adoption in 2026, formalised enterprise platforms in 2027.

Mario then works through the objection Boards most often raise, that enterprise processes need determinism. He argues it covers a narrower slice of work than assumed: soft processes such as drafting and synthesis tolerate nondeterministic output, while hard processes such as compliance and financial reporting are better wrapped with predictive intelligence than replaced. He sets out what minimum lovable governance means for agents: proportionate controls, clear policies on persistent access, monitoring for unauthorised API connections, and an amnesty pathway from shadow use to governed capability. He also flags the energy multiplier: always-on inference removes the overnight lull, and with UK industrial electricity at roughly four times US rates, energy policy becomes AI policy.

This episode is for directors and Boards whose AI policies still address only episodic chatbots. The window is 2026: establish governance now, or discover shadow agentic AI spreading by the second half of the year and retrofit controls on adoption already underway. Read the full article at mariothomas.com

Episode notes

The Invisible Asset: Why Boards Should Govern Data Like It's on the Balance Sheet

Boards steward physical assets with condition checks and ownership, and govern data as if it did not exist. The gap is governance, not technology.

Boards steward buildings, machinery, and vehicles with condition assessments, named owners, maintenance budgets, and impairment tests. Data, which may now contribute more to enterprise value than the buildings that house it, receives none of them. Ownership fragments, quality degrades undetected, and nobody asks whether a dataset is fit for the purposes now being asked of it. A physical asset treated that way would be condemned.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines why data is invisible to governance and what that invisibility costs. Intangible assets now represent 90% of S&P 500 market value, yet IAS 38 prohibits capitalising internally generated intangibles such as databases, so the balance sheet never sees them. Poor data quality costs organisations an average of USD 12.9 million a year, 63% of organisations lack or are uncertain about the data management practices AI requires, and the Institute of Directors reports that a quarter of directors are concerned about the absence of an internal AI policy, strategy, or data governance framework.

Mario sets out the stewardship disciplines Boards already apply to physical assets and extends them to data: condition assessments, a named owner for each strategic dataset, preventive maintenance, impairment testing, and strategic value review. He places the work in the AI Centre of Excellence, reporting to the risk committee, and applies the minimum lovable governance principle: start with the proprietary datasets that matter most, plot them by strategic value against current quality, and govern in proportion to risk. Shadow AI raises the stakes, with 22% of files uploaded to generative AI tools containing sensitive content.

This episode is for directors and the Boards who approve AI budgets without assessing the data those investments depend on. When AI makes a decision, the Board is making that decision. When accounting standards lag, governance must lead. Read the full article at mariothomas.com

Episode notes

The Verification Premium: What Classical Training Reveals About AI Coding Costs

AI coding tools amplify the expertise gap rather than closing it: senior developers capture twice the gains. The verification premium is the cost nobody budgets.

AI coding tools are sold to Boards as a way to reduce dependence on expensive senior developers. The evidence points the other way. Expertise does not become less relevant when AI writes the code; it becomes the factor that decides whether the assistance compounds productivity or compounds debt.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines what his experiment building two applications with Amazon Kiro and Claude Code revealed about where decades of classical software engineering training still matter, and sets it against the research. McKinsey’s analysis found senior developers saved twice as much time as juniors, with gains of up to 55 per cent in greenfield projects falling to 10 to 20 per cent in mature codebases. METR’s randomised controlled trial of 16 experienced open-source developers found a 19 per cent net slowdown, against a predicted 24 per cent speed-up.

Mario then works through the technical debt time bomb: GitClear’s analysis of 211 million changed lines showing an eightfold increase in duplicated code, 70 per cent of it from inexperienced users; the Nature finding that AI models emit up to 19 times more carbon dioxide equivalent than human programmers; the vibe coding cleanup, with more than 8,000 startups facing rebuild costs totalling between 400 million and 4 billion dollars; and the erosion of the junior-to-senior pipeline that creates verification expertise. The governance response is to pair tool deployment with expertise investment, and to ask not whether AI can write code cheaper but whether the organisation can verify that the code creates value rather than debt.

This episode is for directors, CIOs, and the Boards approving AI coding investments on the assumption that tools substitute for expertise. The verification premium is real. The question is whether Boards will invest in it before or after discovering its absence. Read the full article at mariothomas.com

Episode notes

The AI Talent Bifurcation: Are You Building Skills or Collecting Credentials?

Workers with real AI capability command premiums of 28-56%; those collecting credentials without it face a 29% penalty. The same split now reaches the Boardroom.

The AI skills wage data points two ways at once. Workers with genuine AI capability command premiums of twenty-eight to fifty-six per cent, while those who target AI-exposed roles without real capability development face a twenty-nine per cent earnings penalty. The same roles produce opposite outcomes, and the difference is the quality of capability investment, not access to tools.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines what separates premium-earning capability from penalty-suffering credentials, and why the split reaches the boardroom. PwC’s Global AI Jobs Barometer puts the AI skills premium at fifty-six per cent, up from twenty-five per cent a year earlier; Lightcast’s analysis of 1.3 billion job postings finds twenty-eight per cent for one AI skill and forty-three per cent for two or more; Harvard research from 2025 finds the twenty-nine per cent penalty. The Institute of Directors’ NEDs Reimagined paper positions AI competence as a non-executive director responsibility, with Recommendation 11 calling on NEDs to build their understanding of AI.

Mario sets out what genuine capability looks like: verifying AI outputs against domain knowledge, recognising when recommendations do not fit, redesigning workflows rather than inserting tools into unchanged processes, and handling the exceptions automation cannot. He draws on BCG’s 2025 finding that sixty-seven per cent of employees at companies that redesign workflows around AI save over an hour daily, against forty-nine per cent where tools sit inside existing processes, and closes with five questions for the next Board meeting, including whether NEDs can independently evaluate AI strategy or must rely on management interpretation.

This episode is for directors and the Boards who fund AI training. The premium-penalty gap will widen, and the question is which side of it the organisation, and its Board, will be on. Read the full article at mariothomas.com

Episode notes

The Redeployment Dividend: Why AI Will Unleash Your People, Not Replace Them

AI's primary value is not headcount reduction but the intellectual capital it releases from undifferentiated work. Measure only the former and the dividend goes unclaimed.

AI adoption has become synonymous with headcount reduction. Business case discussions are dominated by how many FTEs can be eliminated, and workforce reduction is treated as the primary success metric for AI initiatives, while the intellectual capital trapped in mundane, automatable, undifferentiated work goes unclaimed.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, makes the case for the redeployment dividend: AI’s primary value is not replacing people but releasing trapped intellectual capital, the same people doing more valuable work rather than fewer people doing the same work. He cites Harvard research published in 2025 finding that 25 to 40% of roles are AI retrainable, and its warning that workers targeting high-AI-exposed roles without genuine capability development face a 29% earnings penalty. Deloitte’s 2025 workforce research finds that workers across all age groups prefer mixed human-AI collaboration, and that 60% believe AI can help experienced workers share knowledge.

Mario sets out why the transition matters. He argues for accepting selective atrophy: not every capability humans currently exercise deserves preservation, but strategic reasoning, relationship building, and complex judgement do, and MIT Media Lab research on cognitive atrophy and Bainbridge’s 1983 irony of automation show what is lost when people become passive monitors of AI outputs. He points to the WGA and SAG-AFTRA agreements as evidence that benefit-sharing transitions are achievable when approached as partnership rather than extraction. And he proposes a different success metric, framed through the Well-Advised strategic priorities: measure where freed capacity flows, into innovation, deeper customer relationships, operational resilience, responsible transformation, and genuinely differentiated work rather than cost reduction alone.

This episode is for directors, chief executives, HR and finance leaders, and the Boards whose AI business cases default to layoff arithmetic. The key question is straightforward: are the organisation’s people doing more valuable work, or simply less work? Read the full article at mariothomas.com

Episode notes

Return-to-Work Briefing: Five Forces Reshaping the Board AI Agenda in 2026

Five forces shape the Board's AI agenda in 2026, led by AI embedding into the enterprise faster than it can be governed. None is distant.

AI is embedding itself into the enterprise faster than organisations can govern it, while eroding the human capabilities needed to oversee it. Gartner projects that 40 per cent of enterprise applications will feature AI agents by the end of 2026, up from less than 5 per cent in 2025, and that half of all organisations will introduce AI-free assessments to counter the critical-thinking atrophy AI reliance has created. The luxury of treating AI as tomorrow’s problem has ended.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, delivers a return-to-work briefing on the five forces shaping the Board AI agenda for 2026: AI’s shift from content generation to decision support, inference economics reshaping deployment strategy, embodied AI introducing physical-world liability, verification gaps exposing governance failures, and AI governance professionalising into systematic capability. Among the evidence: Deloitte’s 2026 TMT Predictions expect inference to account for two-thirds of all AI computing power by 2026, and Stanford HAI research finds general-purpose LLMs hallucinate on legal queries between 58 and 82 per cent of the time.

Mario pairs each force with the question a Board should be asking. Is the AI strategy still focused on content generation, or has it pivoted toward decision support? Does the organisation know the carbon footprint of its AI inference? If its AI causes physical harm, who is liable and how is it insured? Has it invested in verification capability proportionate to its deployment? Does the governance structure reflect AI’s strategic importance, or is AI still being treated as a technology project? He closes on what he calls minimum lovable governance: just enough structure to ensure responsible deployment while preserving agility.

This episode is for directors and Boards starting 2026 with AI on the agenda. Boards that defer do not avoid these decisions; they make them by default. Read the full article at mariothomas.com

Episode notes

The Year AI Grew Up: Five Inflections That Changed the Strategic Calculus in 2025

In 2025 Boards stopped asking what AI could do and started treating it as strategic infrastructure investment. Five connected inflections drove that shift.

In 2025 AI stopped being a capability question and became a question of strategic investment in organisational infrastructure. Regulation became enforceable then adapted under geopolitical pressure, energy constraints reached Board agendas, sovereignty fragmented into incompatible ecosystems, the experimentation window closed, and agentic hype met operational reality.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, connects the five inflections of 2025 into a single picture. He traces the EU AI Act’s arc from enforceable prohibitions on 2 February 2025, with fines of up to €35 million or 7% of global turnover, to the Digital Omnibus of 19 November, which proposed deferring high-risk obligations to December 2027 or later. He cites Goldman Sachs’ Powering the AI Era research projecting data centre power demand rising 160% by 2030, and MIT’s finding that 95% of generative AI pilots fail to reach production or deliver measurable ROI.

Mario then works through the implications of each inflection: governance frameworks that must be adaptive rather than static; energy access as a capital allocation decision that determines AI capability; the sovereignty trilemma of trust, speed, or control, where not choosing is choosing; the GenAI Divide, with Menlo Security finding 90% of employees using AI daily outside enterprise controls; and agentic AI, where McKinsey found only 23% of organisations scaling agentic systems. Agentic AI is generative AI in a loop, and the strategic question is where to consciously transfer decision-making authority from people to systems.

This episode is for directors and the Boards entering 2026 with AI on the agenda as infrastructure rather than innovation theatre. The shift demands capital allocation rather than project approvals, long-term commitment rather than pilot funding, and governance as enablement rather than a compliance checkbox. The question for every Board is whether it is ready to treat AI as infrastructure rather than novelty. Read the full article at mariothomas.com

Episode notes

The Return of Traditional AI: Organisations Are Rethinking Their LLM-First Strategies

Forty-two percent of companies abandoned most of their AI initiatives this year, often because generative AI was applied to problems traditional methods solve better.

Forty-two per cent of companies abandoned the majority of their AI initiatives this year, up from seventeen per cent in 2024. The pattern behind the number is consistent: organisations applied generative AI to problems that traditional machine learning or deterministic automation solve better, and they are now recalibrating.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines that recalibration. S&P Global supplies the abandonment figure, Forrester predicts generative AI will orchestrate less than one per cent of core business processes in 2025, the World Quality Report finds sixty per cent of organisations citing hallucination and reliability as top concerns, and BCG’s survey of more than 1,250 firms finds only five per cent achieving AI value at scale. He notes that Salesforce has been building deterministic automation into Agentforce, its Agentforce CTO observing that an LLM given more than about eight instructions starts dropping them. This is maturation, he argues, not failure.

Mario sets out where each approach belongs. Classification against fixed criteria, pattern recognition in structured data such as fraud detection and predictive maintenance, and precision-dependent calculation sit with rules engines and traditional machine learning. Contextual understanding, summarisation, semantic search, and creative generation are natural LLM territory. Hybrid architectures let deterministic systems execute while generative AI supplies context. For Boards, he draws three diagnostics from the Complete AI Framework: problem domain assessment, honest capability matching that includes inference cost at scale, and risk and reliability assessment. He closes with a maturity arbitrage mapped onto the AI Stages of Adoption: proven traditional AI builds confidence at the Experimenting stage, while hybrid architectures serve Adopting and Optimising organisations.

This episode is for directors and Boards facing AI budgets under scrutiny. The question is not whether to invest in AI, but whether each investment deploys the right AI for the specific challenge. Read the full article at mariothomas.com

Episode notes

A New Grid Actor: AI Infrastructure Is Becoming Energy Infrastructure

AI infrastructure operators building their own generation become grid actors rather than consumers, and that changes energy economics, nowhere more sharply than the UK.

The USA faces a 19GW power shortfall by 2028, over 40 per cent of projected data centre demand, and with eight-year grid interconnection queues against 18-month GPU refresh cycles, hyperscalers and AI labs are building their own generation. Capacity built to secure AI workloads typically exceeds what those workloads need, and that surplus turns infrastructure operators from energy consumers into grid actors.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines the scale of the shift and what follows from it. OpenAI’s Stargate facility includes 361MW of on-site generation, Meta’s Prometheus cluster adds 200MW, and combined hyperscaler nuclear commitments now exceed 8GW. Wärtsilä and AVK project European data centre demand rising 250 per cent by 2030, with 40 per cent of existing AI data centres constrained by power availability by 2027.

Mario then works through the progression from consumer to prosumer to grid actor. A 1GW campus at 60 per cent average utilisation holds 400MW of dispatchable capacity off peak, functionally a power station, and that excess can be sold through demand response programmes. For the UK he weighs opportunity against risk: the AI Energy Council, Rolls-Royce SMR’s 470MW units arriving in the mid-2030s, and a 285 TWh system in which a single player at 5 per cent of capacity raises concentration and foreign-ownership questions. He names the emissions trade-off, gas turbines filling the gap until SMRs arrive, and closes with three questions: whether to assess generation potential alongside compute, where backup supply becomes grid export, and what frameworks belong in place before an AI investment makes the organisation an energy market participant.

This episode is for UK Boards and directors who can no longer separate energy strategy from AI strategy. The organisations that anticipate this transition will shape it; those that do not will adapt to rules others write. Read the full article at mariothomas.com

Episode notes

The AI Maturity Mirage: Diagnosing the Gap Between Investment and Readiness

Boards overestimate AI maturity by counting tools and pilots rather than capability. Three patterns create the illusion, and each can be diagnosed before it misleads.

Boards reviewing AI progress see pilots underway, tools adopted across teams, and early efficiency wins. That view misleads, because visible activity bears little relation to genuine organisational capability. The gap between the two is the AI maturity mirage, and it derails transformation strategies before anyone notices.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines the disconnect and the three patterns that produce it. Larridin’s State of Enterprise AI 2025 report finds that 89% of enterprises have adopted AI tools while only 23% can accurately measure their return on investment, and IBM and Ecosystm’s 2025 APAC research shows 85% of organisations claiming data-driven or AI-First status against 11% demonstrating true readiness. The patterns are the tool-centric illusion, where deployments are counted as maturity; the pilot success trap, where isolated wins are read as systemic progress; and hype-driven metrics, where advertising agencies rate AI criticality at 8.1 out of 10 yet embed it in only 16% of operations.

Mario then sets out a three-step diagnostic. Map each function independently to one of the five AI Stages of Adoption, treating pilots that consistently fail to scale within six months as a sign of Experimenting-stage capability. Evaluate balance across the Five Pillars, since tool adoption without governance, people, and value realisation maturity is the signature of the mirage. Test with leading, lagging, and predictive indicators together, because over-reliance on lagging measures inflates perceived maturity. He matches remedies to the gaps: an AI Centre of Excellence with minimum lovable governance, infrastructure coherence before more tools, people programmes, the hub-and-spoke model, and Well-Advised scorecards.

This episode is for directors and the Boards who suspect perceived Optimising status may reflect Experimenting capability. The mirage persists for organisations that mistake activity for capability; for those willing to look clearly, the path to genuine maturity becomes visible. Read the full article at mariothomas.com

Episode notes

Minimum Lovable Governance: The AI Operating Principle Boards Should Use

Minimum lovable governance replaces episodic compliance with continuous, embedded oversight people actually want to use: guardrails that earn adoption rather than enforce it.

AI governance is too often heavy where it should be light and light where it should be heavy: elaborate approval processes for low-risk experiments, minimal oversight of high-stakes autonomous systems, and policy documents nobody reads. The result is friction without assurance. Unloved governance gets routed around, and shadow AI shows how often.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, explains minimum lovable governance, a concept he has referenced across several articles, and why it makes AI governance work. It borrows Eric Ries’s progression from Minimum Viable Product to Minimum Lovable Product and applies it to oversight: the smallest system that achieves the necessary guardrails and that people actually want to use. UpGuard’s November 2025 report found more than 80 per cent of employees, and nearly 90 per cent of security professionals, using unapproved AI tools: governance that exists on paper but fails to govern.

Mario sets out what lovable means operationally: embedded in how work happens, continuous rather than episodic, proportionate to risk, and clear at the point of decision. He explains why the approach has only recently become viable, pointing to the EU AI Act’s risk-tiered architecture, ISO 42001 and ISO/IEC 42006:2025 as reference points, and AI-assisted governance making continuous oversight achievable while accountability stays with humans. He then works through five principles, each with a Board test question, such as whether an auditor calling tomorrow would be answered in seconds or weeks, and whether a marketing chatbot goes through the same approval process as a credit decisioning system.

This episode is for directors, chief executives, and the Boards whose governance reports say everything is under control while most of the organisation works outside it. The choice is to build governance people route around, or governance people want to use, and the outcomes will differ accordingly. Read the full article at mariothomas.com

Episode notes

World Models: The Next Horizon in AI for Predictive Enterprise Intelligence

World models move AI from reacting to anticipating: systems that simulate future scenarios, with aviation and finance already seeing the operational gains.

Current AI excels at analysing what has happened. World models, an emerging approach, build internal simulations of how reality works and predict what will happen next, moving AI from recognising patterns to anticipating futures. For Boards already tracking leading and lagging indicators, they promise the technical realisation of the predictive indicator.

In this episode of The Board in the Machine, Mario Thomas, Chartered Director and Fellow of the Institute of Directors, examines world models as the next horizon in predictive enterprise intelligence, drawing on Yann LeCun’s work on the Joint Embedding Predictive Architecture and on the gains current predictive AI already delivers: aviation pioneers reporting twenty to thirty per cent improvements in operational efficiency in BCG’s AI-First Airline report, and PwC’s 2025 finding that industries exposed to AI disruption achieve three times higher revenue per employee growth. He is candid about the barriers, which LeCun himself acknowledges as huge practical impediments: computational demands that may need quantum breakthroughs, vast real-time data requirements, and energy consumption that could make large-scale world models economically unviable.

Mario sets out grounded timelines, with narrow-domain world models perhaps two to three years away and enterprise-ready systems five to seven, and argues for a dual strategy: build foundational predictive capabilities with current AI now whilst monitoring world model developments for strategic timing. He imagines the transformation across aviation, financial services and manufacturing, then turns to governance: the ethical boundaries Boards must set when a simulation reveals a profitable but harmful path, the case for minimum lovable governance, transparency about the scenarios and assumptions behind each prediction, and risk management that simulates whole consequence chains rather than single events.

This episode is for directors and Boards preparing for predictive intelligence. Which operational predictions would transform the organisation’s competitive position when world models arrive? That answer determines where to build foundations now. Read the full article at mariothomas.com