How AI is changing the way boards evaluate strategic options

AI is changing strategic evaluation at board level by giving directors faster, more targeted access to the substance behind management's proposals. It does not change who is accountable for evaluating them. 

Strategy papers have grown longer and more data-heavy over the past decade, and management teams now bring considerably more analytical capacity to a recommendation than the board has time to absorb in a pre-read. The result is a widening gap between what management knows about a proposal and what the board can reasonably interrogate before approving it. Spencer Stuart's most recent board index found that only 43% of CEOs believe their board directors have specific subject-matter expertise aligned with the company's most pressing issues, against 63% of directors who believe they do. That gap in perceived readiness sits close to the centre of the strategic evaluation problem boards now face. 

AI offers a partial answer. Used well, it narrows the distance between the volume of information a strategic decision generates and the amount a board member can genuinely engage with before the meeting. Used poorly, it adds another layer of material to get through. This article sets out where that distinction holds, and where it does not. 

Where AI is changing the picture 

AI is affecting strategic evaluation in three distinct, observable ways: how board members synthesise complex materials before a meeting, how they interrogate the assumptions behind management's proposals, and how they track whether previously approved strategies are being delivered. None of these is theoretical. In PwC's 2025 Annual Corporate Directors Survey, 35% of directors said their boards had already integrated AI, including generative AI, into their oversight activities, a figure expected to keep rising as the tools mature. Each of the three areas below already changes a specific part of the board's preparation and oversight workflow. 

Faster synthesis of complex information 

Natural-language querying (asking a system a plain-English question about a document rather than searching or reading it in full) changes what is practical to do with a large board pack in a limited amount of preparation time. 

A board member preparing for a strategy discussion might receive 200 pages of supporting material: a strategic rationale, a financial model, market analysis, risk assessment, and management's recommendation. Without AI assistance, the realistic choices are to read everything, rarely achievable given the time most directors have; to rely on the executive summary, which loses nuance and the assumptions worth challenging; or to arrive under-prepared. AI-assisted synthesis adds a fourth option: targeted interrogation of the document itself. A director can ask what the three biggest assumptions in the financial model are, or how a proposal compares with what the board approved in 2023, and receive an answer drawn from the materials rather than from memory or someone else's summary. 

Faster synthesis does not mean faster judgement. It means better-prepared judgement.

Scenario modelling and risk analysis 

AI does not let boards build management's models themselves, but it does let directors ask sharper, more specific questions about the assumptions inside those models. 

Part of the board's job in strategic evaluation is to challenge the assumptions behind a recommendation, not just the recommendation itself. AI tools allow directors to probe those assumptions more systematically: what the model looks like if revenue growth runs three percentage points lower than forecast, or what comparable transactions look like for a deal of this type and size. This narrows, without eliminating, the information asymmetry between a management team that has been modelling a proposal for months and a board reviewing it in a single pre-read. 

The shift in attitude towards AI-assisted analysis in high-stakes decisions is measurable outside the boardroom too. A 2026 FT Longitude survey of 1,000 dealmakers commissioned by Datasite found that 62% now believe human-only decision-making in complex decisions is indefensible. The same research found that accuracy (71%), security (70%), and reliability (58%) are the attributes experienced decision-makers most require of an AI tool working on high-stakes material, a useful benchmark for what boards should expect of any system operating on board papers. 

AI can surface better questions. It cannot resolve them, and boards should be wary of any framing that suggests otherwise. 

Monitoring strategic progress in real time 

Boards have traditionally depended on management for updates on how an approved strategy is progressing, which means their view of execution is episodic rather than continuous. 

AI-assisted monitoring closes part of that gap. Structured tracking of commitments, decisions, and milestones gives the board a running record of what was agreed, by whom, and against what timeline, rather than a fresh narrative update at each meeting. This matters for the accountability question raised later in this article: a board that approved a strategic initiative 18 months ago should be able to see, without relying entirely on management's account, whether the conditions attached to that approval have been met. 

Where the board's judgement still matters most 

AI changes how much information a board can access and how quickly. It does not change the nature of independent scrutiny, the ability to read dynamics in the room, or who is accountable for the decision that follows. 

At least three areas remain irreducibly human. Independent scrutiny is the board's core function: AI can help a director prepare sharper questions, but it cannot supply independence, and the judgement call of whether an initiative is the right one, whether management is the right team to deliver it, or whether the risk is acceptable, remains a human decision. 

Reading the room is a second area. Strategic discussions often turn on dynamics no AI system can observe: a management team that is overconfident, a financial model that is technically sound but built on optimistic assumptions, a chair who senses dissent is being suppressed before anyone says so. These call for experienced judgement, not analysis. 

Accountability is the third, and the one with the clearest practical consequence. When a board approves a strategic decision, it accepts responsibility for that decision, and that responsibility cannot be delegated to a tool. The board must be able to explain its reasoning and stand behind it. 

AI changes what information the board can access. It does not change who is accountable for the decisions that follow.

This is consistent with where the wider professional consensus on AI-assisted decision-making currently sits. The same FT Longitude/Datasite research found that only 22% of dealmakers are willing to follow an AI-generated recommendation on whether to sign a deal, even though most now use AI-assisted analysis somewhere in the process. Experienced decision-makers draw a clear line between AI-assisted analysis and AI-led decisions. Boards should draw the same one. 

How to embed AI into board-level strategic discussions responsibly 

Responsible adoption of AI at board level starts with where the technology is applied, not only how it is secured. 

The lowest-risk, highest-value application is pre-meeting preparation. Document synthesis, question formulation, and background research are appropriate starting points; AI should not be generating draft resolutions or recommending whether a proposal should be approved. Boards adopting AI should set out, in a clear policy, which tools are approved, what data they can operate on, and what their outputs can inform rather than determine.

Few currently do: EY's review of Fortune 100 proxy disclosures found that, across the first eleven months of 2025, only 12% of companies disclosed that board members had received any education or training on AI. It also helps to know whether preparation has actually happened, not just whether the tools were available. A preparedness signal, showing whether board members have engaged with the materials before a discussion begins, gives the chair something concrete to act on rather than discovering gaps in the room. 

Security is not a separate consideration bolted onto this policy; it is part of what makes the policy credible. Board papers contain market-sensitive information, commercially confidential data, and personal information about directors and executives, and any AI tool operating on that material should meet the same standard as the rest of the board's technology stack. 

Boardroom AI security: what to look for

Any AI tool used in board-level governance should operate on a zero-training-on-customer-data principle, provide full audit trails of AI interactions, apply role-based permissions so board materials remain accessible only to authorised users, and comply with applicable data residency requirements. These are not optional features; they are governance requirements.

The same discipline applies to the board's own record. AI-assisted minutes and decision records are only as reliable as the governance framework around them. Research from the Chartered Governance Institute UK & Ireland found that 74% of governance professionals are concerned about the accuracy of AI-generated content in corporate reporting. The company secretary's professional judgement, not the tool, should remain the final word on what the record says. 

How Sherpany's AI capabilities support strategic oversight 

Sherpany's AI capabilities were built for the governance environment boards actually operate in: security-first, focused on preparation rather than decisions, and human-in-the-loop throughout. 

• Document Copilot lets board members query strategy documents and board papers in natural language and receive concise, contextualised answers drawn from the materials themselves. No board data is used to train the underlying model, and every interaction is logged in an immutable audit trail. 
• Minutes Copilot integrates with Microsoft Teams, Zoom, Webex, and Google Meet to produce AI-assisted documentation of strategic decisions, with accuracy validated by the company secretary before any record is finalised. 
• Topic Hub structures strategic topics before they reach the agenda, anchoring the board's AI-assisted preparation in a consistent view of the organisation's strategic priorities, rather than a fresh set of papers each cycle. 

All three sit on the same security foundation as the rest of the platform: Swiss data residency, AES-256 encryption at rest and in transit, ISO 27001 and SOC 2 Type II certification, role-based permissions, and no training of AI models on customer data. 

AI gives boards a genuine opportunity to close part of the information gap with management and to prepare more substantively for strategic discussions. It does not change the fundamental nature of what boards do. 

The quality of a board's strategic oversight still depends on the independence, experience, and judgement of its members, and AI does not substitute for any of these. Used well, it is a tool that makes those qualities more effective, freeing time spent assembling information for the harder work of interrogating it. Used poorly, or adopted without a clear policy on what it is permitted to do, it adds complexity without adding insight. 

The boards that get the most from AI in strategic evaluation will be the ones that treat it as exactly that: a tool for better-prepared judgement, not a substitute for the judgement itself.