The question boards think they are answering
Most boards believe they are struggling with AI.
The latest Board Value Index suggests they are struggling with something else entirely.
They ask which tools to buy and how much to hand over. Every question points one way: how do we bring AI in?
A 2026 study from Board Intelligence, one of Europe's largest board technology and advisory firms, suggests the question is misdirected. The firm published its Board Value Index in June 2026, surveying more than 400 non-executive directors, CEOs, and CFOs at companies with over £50 million in turnover across the UK, the US, the Nordics, and the Middle East.
The finding that matters has nothing to do with AI.
Boards can no longer assume that their existing judgment structures remain adequate.
This starts to bite as AI systems and autonomous agents move into the decision path. Judgment stops being a single act by a single person and becomes a process spread across models, agents, executives, and committees. Once judgment spreads across layers, no one can point to who holds authority or who answers for the result. A board that never designed those answers inherits them by default. The Board Value Index shows organizations discovering that the defaults no longer hold.
What the survey actually found
The heaviest number in the Board Value Index has nothing to do with artificial intelligence.
Eighty-six percent of directors said rigid or inconsistent decision-making frameworks had contributed to a delayed, rushed, or poor decision in the previous six months. This says nothing about markets, competitors, or regulation. Directors admit that the way their boards reach decisions produces bad outcomes.
The directors named specific causes: the decision-making framework or process itself (34%), the unclear division of roles between the board, executives, and committees (32%), and the quality of information reaching the board (29%).
Frameworks, roles, information. Directors sitting in the boardroom named these three as the obstacles to fast, sound decisions.
AI did not create any of these problems. Each would exist in a boardroom that had never seen a model.
The question boards cannot yet answer
The survey also captured how boards respond to AI.
Eighty-four percent of boards reported that they have begun debating which decisions should stay human-led and which should go to AI. On its surface, this looks like progress.
Say the question out loud and a problem appears. What counts as a "human-led decision"? Where does an "AI-led decision" begin and end? Boards debate which way to lean before drawing the line they lean across.
So an 84% engagement rate produces little resolution. The interest is real. The question cannot yet be answered in the form boards ask it. Many boards debate a boundary that none of them has drawn.
Why AI Governance, AI Ethics, and AI Risk miss the center
Most commentary reaches for a familiar vocabulary: AI Governance, AI Ethics, AI Risk.
Each is legitimate. Each addresses a real requirement: how to oversee AI, how to constrain it, how to contain its downside. None is wrong.
All three sit to the side of what the Board Value Index exposed. The survey did not describe a failure to supervise AI. It pointed one level earlier, before AI enters: the structure of judgment itself.
The real issue is judgment
Look again at the three obstacles Board Intelligence identified: decision-making frameworks, roles and responsibilities, and information quality.
These are not three problems. They are three parts of one thing.
A decision-making framework structures a judgment. Roles and responsibilities decide who holds it. Information quality decides what enters it. Structure, ownership, input: each is a component of one act, judgment.
The Board Value Index reads on its surface as a list of governance weaknesses. Read it again and it becomes an inventory of the parts of judgment that no one has assembled into a whole.
Board Intelligence reaches this diagnosis without naming it. The report lists the components of judgment. It never treats judgment as something a board can design.
Boards see the change coming and expect to stay the same
The sharpest tension in the data sits between what boards expect of the world and what they expect of themselves.
Directors accept that AI and technological disruption will reshape the organizations they oversee. Yet 40% believe their own boards will need little or only incremental change over the next five years. They expect the enterprise to transform while assuming the body that governs it can stay as it is.
Two more numbers deepen the picture. Only 37% of directors regard their board as essential to value creation, so nearly two-thirds do not see the board as a source of value. And only 18% say their board strongly enables innovation, a low figure for bodies meant to steward long-term growth.
Attention points backward too. Forty-one percent of directors spend at least half of their meeting time reviewing past performance rather than shaping future strategy. A board that spends half its hours looking back, using frameworks built for an earlier era, reproduces the judgments of that era.
Put together, the numbers describe one gap. Directors admit their judgment is flawed. They do not expect to redesign themselves. They doubt they create value. The problem is not confidence. No one has treated the board's judgment as something to build.
Boards have governance. They still cannot say who should decide.
Consider the word we reach for by reflex: governance.
Board effectiveness, oversight mechanisms, the allocation of authority and responsibility all count as governance, and all of it matters. But notice what governance designs. It designs who supervises whom. Who monitors, who reports, who answers in the end.
Governance answers oversight. It leaves judgment unanswered.
Ask a board a concrete question and the gap appears. Who decides this investment, a person or an AI system? At what layer does the authority to halt a contract sit? How far may a machine's output stand in for a human evaluation before the decision stops belonging to the human?
A board can usually produce an org chart of oversight. It rarely has an equivalent structure for judgment. The supervisory architecture runs to detail. The judgment architecture stays blank.
That blank space is where the 86% live. The discomfort those directors report is not a failure of oversight. It is the missing structure for who decides what, and where.
Designing judgment
Organizations have long treated judgment as personal. An experienced individual, deciding in the moment. A matter of talent, not structure.
The survey points another way. When 86% of directors report failing decision structures, and 84% debate where the line between human and machine judgment should fall, judgment stops being a question of individual capability. It becomes a question of organizational structure.
A board can design anything structural.
This is the premise behind Decision Design: the act of judgment is a legitimate object of design.
Decision Design does not improve decisions alone; it designs the authority structure within which decisions become institutionally legitimate.
Decision Design begins before a decision is made.
Making a better call on a single investment improves a decision. Deciding in advance who holds the authority to make that call, on what basis, and where a machine's contribution ends, designs the structure in which the call becomes legitimate. The first is judgment quality. The second is judgment architecture. Decision Design builds the second.
Decision Design sets the structure through which a board reaches a judgment: who frames the question, who generates the options, who decides, and who bears the consequences. It fixes that arrangement in advance.
Decision Design does not get individual answers right. It does not predict which investment will pay off or which hire will succeed. That is the quality of a decision, not its structure.
It addresses the gap the Board Value Index exposed: in most organizations, no one has designed who decides how much, or where authority passes from human to machine. The 86% did not misjudge the content of their decisions. They worked without a designed structure for judgment. Decision Design makes that structure the primary object.
Decision Boundaries: where authority changes hands
One concept sits at the center of Decision Design: the Decision Boundary.
A Decision Boundary defines where legitimate authority begins, ends, transfers, or is reclaimed.
An operational threshold is a number that triggers a workflow, an amount above which a second signature is required. A Decision Boundary works differently. It names the legitimate author of a judgment, and it sits before any workflow begins.
A Decision Boundary defines, for a given judgment, the range a human keeps and the range delegated to a machine or a lower layer. It sets how far a decision follows an AI system's output, and the point at which a human takes the decision back. It draws one explicit line.
An approval flow routes an already-defined item through a sequence of signatures. A Decision Boundary decides who authored the judgment in the first place, so it comes before the flow.
A Decision Boundary solves the problem behind the 84%. Boards debating "human or AI" without resolution argue over direction without drawing a line. A Decision Boundary changes the question from "which way should we lean" to "where does the line fall." The first has no stable answer. A board can design, document, and revise the second.
Decision Logs: keeping accountability continuous
Drawing a boundary does not help if no one can later reconstruct how a distributed judgment was made. When decisions pass across humans, models, and agents, accountability dissolves at the seams. A Decision Log addresses this.
A Decision Log preserves accountability across a distributed judgment, not the outputs alone.
An ordinary audit log records what happened: this output appeared, this action ran, at this time. That is necessary and not enough. In an AI-augmented organization, a model recommends, an agent filters, a manager adjusts, and a committee ratifies a single judgment. If each layer records only its own output, the chain of responsibility fragments. No one can say where authority rested when the board made the judgment.
A Decision Log records the outputs and the boundary in force: which layer held authority, what the machine contributed, where the human took over, and who ratified the result. Accountability means nothing if no one can trace it through the point where judgment split across layers. An audit log tells you what the system did. A Decision Log tells you who was responsible, and why that responsibility was legitimate.
This matters under scrutiny. When a regulator, a board committee, or a court later asks why the board made a decision, an audit trail of model outputs cannot answer the question that counts: who exercised authority, and was that authority theirs to exercise? A Decision Log answers it by keeping the boundary alongside the outcome. In an organization where judgment spreads across AI agents and human layers, a Decision Log decides whether accountability survives the handoffs or vanishes at each one. Without Decision Logs, Decision Boundaries cannot remain institutionally auditable.
Why existing frameworks are necessary but not sufficient
Decision Design does not replace the concepts organizations already rely on. It occupies a layer they leave unmodeled. These frameworks are not flawed. Each has a defined scope, and that scope stops short of institutional judgment.
Governance designs the structure of oversight. It sets who supervises whom and who answers to whom. It matters, and it acts after the judgment, at the point of review. It does not name who holds the judgment.
Digital transformation redesigns work. It moves processes from paper to systems, from manual steps to automated ones. It changes how people do the work. It does not name the author of a decision.
Automation designs execution. It runs defined tasks at speed and without variation. It works within its scope. But what to decide, and who decides it, sit upstream of execution.
AI Ethics designs norms. It sets the principles an AI system should uphold. It matters. It does not name which decisions those principles govern, or the point at which a human must take authority back.
Oversight, work, execution, norms: four distinct layers, each necessary, none of them designing who holds institutional judgment. That layer belongs to none of the four. Decision Design fills it and complements the others rather than competing with them. A well-governed, digitally mature, heavily automated, ethically principled organization can still lack any designed answer to who decides.
Adding more of the existing layers does not close the gap. Stronger oversight supervises judgments with more rigor, yet it does not name who should author them. More automation executes decisions faster, yet speed on an undesigned boundary spreads the ambiguity wider. The layers reinforce one another once someone designs the judgment beneath them. Without that layer, each new investment refines the parts of a structure whose center no one drew.
What this looks like in practice
Decision Design is not an abstraction. A board draws a Decision Boundary in a specific process, at a specific place. Several cases show where a board can set the line on purpose.
Board investment approvals. Delegate the quantitative evaluation to AI; keep the risk appetite and the final commitment with humans. The line sits between producing the numbers and choosing to stake the organization on them.
M&A governance. Lean target screening and first-pass due diligence on AI; keep valuation and integration judgment with people. The line sits between collecting the facts and interpreting them.
Enterprise AI agents. Humans set the range within which an agent may act on its own, in advance. Here the line runs inside the agent's own authority, and it must be fixed before execution, not negotiated after.
Procurement approval. Let AI handle routine, low-value requests; escalate exceptions and high-value cases to humans. The nature and materiality of the request draw the line.
Contract review. Let AI check standard clauses; keep acceptance of non-standard clauses a human judgment. The line sits between what is standard and what is not.
HR evaluation. Delegate the aggregation of performance data to AI; keep the evaluation and its accountability with humans. The line sits between aggregation and evaluation, and a machine's output is not the evaluation.
Risk committees. Delegate signal detection to AI; keep the judgment of whether a signal is material with humans. The line sits between detection and significance.
Executive escalation. Define in advance the conditions under which a decision must rise to a more senior layer. The line sits between what a layer resolves on its own and what it hands upward.
The line falls in a different place each time. The question stays the same: how far do we delegate, and where do we take authority back? Decision Boundaries make that line explicit, process by process. Decision Logs keep it accountable once decisions move across humans and machines.
The problem was never AI, and never only the board
Return to where the Board Value Index began. The problem it uncovered is not, at root, a problem of the board.
The board is the first place the gap showed. Its entire function is judgment, so the absence of designed judgment shows up there first and sharpest.
Nor is it a problem of AI adoption. AI is the light that fell on the empty space, not the space itself. The discomfort sat there all along; capable machines made it impossible to ignore.
The real problem sits further upstream. Organizations have rarely designed judgment. Who decides how much, and where authority passes from one actor to another, has been left to habit and individual discretion, unexamined and unstated.
So the survey's numbers cohere. The 86% who report failing decision structures, the 84% debating a line they have not drawn, the 40% who expect not to change, the 37% unsure of their own value, the 18% who feel they enable innovation, the 41% still facing backward: not six separate findings. Six views of one absence.
Organizations have designed governance. They have designed processes. They have designed accountability. They have almost never designed judgment.
Decision Design provides the missing judgment architecture. The Decision Boundary is the line at its center. The Decision Log keeps that line accountable once judgment spreads across humans and machines.
The Board Value Index was never a report about artificial intelligence. It was a report about the institutional design of judgment, and a signal that the layer where that design belongs has, until now, stayed blank.
Decision Design is a judgment architecture framework proposed by Ryoji Morii, founder of Insynergy Inc., for structuring authority, accountability, and decision boundaries in AI-augmented organizations.