How AI Scenario Modelling Supports Better Board Decisions

Two years ago, a board's risk committee might well have spent an afternoon arguing over three scenarios for the year ahead: a base case, a downside, an optimistic stretch goal. Now AI can generate dozens of variations in roughly the time it used to take just to agree on the first three. 

AI improves scenario planning by widening the number and rigour of futures a board can consider, not by making any single prediction more accurate. As Wendi Backler, Alan Iny and Moe Turner of the BCG Henderson Institute put it in Harvard Business Review, “it’s human nature to want a crystal ball,” especially when uncertainty is running high. 

AI can't supply one. What it can do is help a board hold several plausible futures in view at once, rather than anchoring on whichever one feels most likely. 

That gap matters because most boards aren't there yet. Fewer than 40 per cent of directors believe their organisation has an adequate strategy for managing geopolitical risk, according to director research cited through the Harvard Law School Forum on Corporate Governance. Scenario modelling won't predict which shock lands first; it can make sure the board has already thought through how several different ones would play out, and what the company would do about each. 

In this article: 

• Why scenario planning has become a standing board capability, not an occasional workshop. 
• How AI expands the number and rigour of scenarios boards can consider. 
• Why the boardroom debate around a scenario still matters more than the model that produced it. 
• A short test for whether your board's scenario planning is shaping decisions. 

Why scenario planning has become a board-level capability 

Scenario planning's role in board governance is to pressure-test a strategy before the world does it instead. Every strategic plan, however well-researched, embeds a bet on how the future unfolds: which markets grow, which regulations hold, which competitors stay put. Scenario planning forces the board to ask what happens to that bet under different conditions, and whether the company would survive, or even benefit, if the world moved differently than management currently expects. 

This has shifted from a periodic exercise to something closer to a standing capability for one straightforward reason: the range of plausible futures has widened. Trade policy, interest rates, energy costs and the pace of AI adoption itself are all moving at once, in combinations that are harder to reason through informally than they were five years ago. A board that tests its strategy against a single future is, in effect, choosing to be surprised by whichever of the others arrives. The broader shift from periodic to continuous strategic monitoring, of which scenario work is one part, is covered in our companion piece on continuous strategic adaptation; the focus here is on how the scenarios themselves get built and tested. 

How AI expands strategic foresight through scenario modelling 

Boards use AI for strategic foresight in roughly three ways: generating a wider set of plausible futures than a workshop could produce by hand, stress-testing the assumptions underneath the current strategy, and surfacing risks or openings that would otherwise stay buried in data nobody had time to read closely. None of the three works well if the board treats the output as a finished answer. AI can generate scenarios fast; it can't decide which ones deserve the board's attention, or what the company should do about them. 

Exploring multiple plausible futures 

A board historically limited itself to a handful of scenarios because building each one took real analyst time: gathering data, running the numbers, writing up the narrative. 

AI collapses that cost. A model can generate dozens of internally consistent variations, combining different assumptions about interest rates, competitor moves, regulation and demand. 

More scenarios are not automatically more useful, though. A board drowning in forty AI-generated futures isn't better off than one weighing three well-chosen ones; it's simply overwhelmed in a new way. What AI adds is coverage: making sure the scenarios a board eventually discusses were drawn from a wide set, rather than the three someone happened to think of in a planning workshop, with the judgement about which ones matter still sitting with the people in the room. 

Stress-testing strategic assumptions 

Every strategy rests on assumptions the board rarely revisits once a plan is approved: that a key input cost stays roughly stable, that a regulatory framework holds, that a competitor keeps behaving the way it always has. AI is useful here because it can hold those assumptions against live data continuously, rather than waiting for the next scheduled review to notice one has quietly stopped being true. 

KPMG's John Rodi, Anne Zavarella and Patrick Lee frame this as one of the core disciplines boards need in 2026: “stress testing strategic assumptions,” alongside analysing downside scenarios and bringing in outside perspective, in a memorandum published through the Harvard Law School Forum on Corporate Governance. The discipline matters more than the tool. A board that stress-tests its assumptions once a year, with or without AI, is still working from a stale picture most of the time. 

Identifying emerging risks and opportunities 

Eighty-two per cent of CFOs report moderate to high exposure to trade-policy disruption, yet only 29 per cent feel confident in their current forecasting models, according to Gartner's 2025 CFO research. That gap between exposure and confidence is exactly where AI-assisted scenario modelling earns its place: not by closing it through better prediction, but by giving the board and management a shared, continuously updated view of where the exposure sits. 

Applied well, AI surfaces the risk or opportunity a board would otherwise catch too late: a supply chain concentration that only shows up when three scenarios are compared side by side, or a market opening that only becomes visible once a downside case forces the question of where the company would look for growth instead. The value isn't the individual scenario; it's the comparison across several at once, an exercise humans are slow to do properly under time pressure. 

Identifying the risk is only part of the exercise. Simon Laffin, former chairman of Flybe, argues that boards should think beyond risk prevention to what they would actually do if a risk materialised. As he puts it on the Boardroom Confidential podcast: “Too much time is spent apparently preventing a problem, and not enough is spent on: if this starts to fail, what do we do? Because the moment you start thinking about that, you think — I need to be monitoring this closely. You're thinking ahead (...)” 

That distinction matters for scenario modelling. The strongest scenarios do more than show the board different versions of the future; they help define the signals to watch and the actions each outcome would trigger. 

Why governance remains essential 

AI can generate the scenarios; it can't decide which one the board should worry about, or what the company should do if a downside case starts looking like the base case. That judgement is the part of the job that doesn't delegate. 

Guy Gecht, board chair at Logitech, made a related point to McKinsey when discussing AI oversight more broadly: “the board’s role is not to manage AI.” His point was about tempo and boundaries: boards need to let the organisation move at the pace the opportunity requires, while making sure someone has drawn a clear line around what must never be allowed to go wrong. Applied to scenario work, that means treating an AI-generated scenario as an input to a boardroom argument, not a substitute for having the argument. 

In practice, that argument goes better when a few basics are in place. Scenarios need to reach the board early enough to be read properly, not summarised for the first time in the meeting. Directors get more out of a scenario discussion when they have already formed a view of which assumptions worry them most, rather than forming that view live in the room. And when a scenario does change the board's thinking, between scheduled sessions if the shift is significant enough, the reasoning behind it deserves the same record-keeping as any other significant decision, not an informal note nobody can find again in eighteen months. 

The test of a good scenario-planning board 

Three questions cut through most of the noise. Could the board name the two or three scenarios it is currently watching most closely, and explain why those and not others? Has a scenario changed a decision in the last year, or do they get discussed and then filed away? Would a new director be able to see, from the record, which assumptions the board most recently stress-tested and what it concluded? A board that struggles with any of these is treating scenario planning as a presentation to sit through, not a tool that shapes what it decides. 

How Sherpany supports AI-assisted strategic deliberation 

AI-generated scenarios are only useful to a board if the deliberation around them is easy to find again, not scattered across email threads and slide decks that get overwritten at the next update. 

Sherpany's Topic Hub keeps a strategic theme, and the scenarios tested against it, in one place over time, so a board revisiting an assumption six months later starts from the actual record rather than someone's memory of the discussion. Comments let directors challenge a specific scenario or assumption directly against the material itself, capturing the disagreement in context rather than losing it in a separate email chain. The Document Management Library holds the scenarios, the source data behind them and the board's own analysis together, securely, and accessible to directors preparing individually before a session. 

For a fuller framework on building continuous, AI-assisted strategic oversight, our guide, 5 Practical Ways to Build an AI-Ready Board, covers the underlying governance habits in more depth. 

The Point Was Never to Predict Correctly 

No scenario, AI-generated or otherwise, tells a board what will happen. What a good set of scenarios does is make sure the board has already argued through several different versions of the future before committing to a strategy built on just one of them. 

AI changes the economics of that exercise: more scenarios, built faster, tested more often. It doesn't change who is responsible for deciding which ones matter, or for living with the consequences of the bet the board eventually makes. Boards that get real value from scenario modelling treat it as a standing discipline that shapes live decisions, not a workshop output that gets filed away once the meeting ends. 

If you would like to see how Sherpany supports AI-assisted scenario planning and strategic deliberation at board level, book a free consultation today and find out how Sherpany can help.