AI and the viable firm

Are you governing AI for performance—and viability?

Seven questions for boards when the answers are still being discovered.

AI is already changing productivity, but the board’s question is larger. It may alter what customers value, where competitive advantage sits, how the organisation works, how quickly strategic assumptions expire, and how close unfamiliar systems come to consequential decisions.

These questions are not a test of whether the company has “the answer”. They examine whether the board and management are governing the process of discovery with sufficient evidence, challenge and adaptability.

01

Value and advantage

If customers can do more for themselves and competitors can access similar AI, what will customers still value from us—and why will we capture the economics?

Why this belongs at board level

AI may democratise capabilities on which existing propositions and margins depend. The strategic question begins with changes in customer behaviour, market boundaries and bargaining power—not with an internal list of use cases.

Evidence worth asking for

Observed changes in customer behaviour and willingness to pay; competitor, entrant and rebundling scenarios; defensible complementary assets; and economics measured per completed customer outcome, not per task or token.

02

Where AI lands

Where in the customer-to-outcome value chain does AI materially change the work—and how much authority are we giving the resulting system?

Why this belongs at board level

‘AI’ covers systems operating in different domains and with very different levels of consequence. A system that informs a colleague is different from one that recommends, decides or acts close to customers, money, employment or physical assets.

Evidence worth asking for

End-to-end value-chain maps rather than functional use-case lists; distinctions among systems that inform, recommend, decide or act; proximity to consequential outcomes; and clarity about the model, data, harness, tools, permissions and suppliers involved.

03

Operating coherence

When AI accelerates one activity, where does the constraint move—and where do people remain the interface to a world the model does not contain?

Why this belongs at board level

Local productivity does not guarantee improved system performance. Faster coding can create a testing queue; faster analysis can increase the burden on judgement and decisions. In the real economy, people often remain the interface between digital outputs and physical, social or institutional context.

Evidence worth asking for

End-to-end performance rather than isolated productivity; effects on QA, testing, decisions, assurance and exception handling; clearly defined human judgement points; and plans to preserve the formative work through which future expertise develops.

04

Time, horizons and options

As AI shortens the useful life of strategic assumptions—while assets, capabilities and controls may still take years to build—are our strategy and investment horizons adapting? Which commitments should be staged or kept reversible?

Why this belongs at board level

Technology and competitive cycles may move faster while investment, regulation, physical assets, skill formation and organisational change remain slower. The board must govern decisions across these different clocks without treating every possibility as equally urgent or equally certain.

Evidence worth asking for

Different decision and investment approaches for the core business, emerging opportunities and exploratory options; patient investment in durable foundations; explicit costs of waiting and committing; and observable triggers to scale, pause, switch or stop.

05

Risk, proximity and assurance

How is AI changing the speed, scale, concentration and proximity of risks the board already oversees—and are appetite, controls and assurance designed around what the system can actually do, rather than simply which model it uses?

Why this belongs at board level

AI is not only a new category on a technology risk register. It can change strategic, operational, conduct, cyber, people, supplier and reputational risks—and reduce the distance between an error and its consequence. The model is only one part of the system that creates the outcome.

Evidence worth asking for

AI treated as a modifier of enterprise risks; exposure tiered by authority, consequence and reversibility; aggregate dependencies on common models, data and providers; known model constraints; and independent testing of the whole deployed system in the context in which it is used.

06

The assumption at risk

Which assumption about AI would most alter our strategy if it proved wrong—and what observable evidence would disconfirm it?

Why this belongs at board level

When the useful life of assumptions is shortening, a confident forecast is less valuable than knowing which assumptions carry the strategy and how the board will recognise when one has expired.

Evidence worth asking for

A consequential assumption stated precisely; an accountable owner; evidence on both sides; a review date; and a pre-agreed decision, stopping or escalation trigger.

07

The recurring board question

What has management learned that changed its previous view—and what evidence would cause it to change again?

Why this belongs at board level

In a period of genuine experimentation, the quality of learning is part of governance. Activity is not the same as learning, and learning is not valuable unless it changes a decision.

Evidence worth asking for

Decisions changed, investments stopped, options exercised and controls redesigned—not simply pilots launched, productivity reported or examples of successful adoption.

The board does not need management to predict the final shape of AI.

It should expect management to identify consequential assumptions, gather relevant evidence, understand the whole system being deployed, and change course when experience challenges the original view.

Not knowing is not necessarily a failure. Not knowing what would allow the company to learn—and what it would do next—may be.

Read the articleExplore a facilitated board diagnostic

This resource is a prompt for board inquiry. It is not an assurance opinion, risk assessment, legal interpretation or substitute for reviewing management evidence in the company’s specific context.

© 2026 Better Business Design Consulting Ltd. All rights reserved.