The Board Decision Gap: What 1,000 Dealmakers Tell Us About AI and Governance

The question of how boards should govern AI is no longer theoretical. AI is already shaping the analysis, the recommendations, and increasingly the decisions that define how organisations compete. For boards, the task is not to take a position on whether AI belongs in high-stakes decision processes. That debate has largely been settled. The task is to determine what governance belongs around it. 

In March 2026, Datasite commissioned FT Longitude to survey 1,000 dealmakers across the Americas, EMEA, and APAC. Sherpany contributed to the project. While the research focuses on M&A, its findings about AI adoption, accountability, and board-level decision quality speak to a much broader set of organisations. 

You can download the full report here. 

Where AI has taken hold 

Across the most intensive stages of complex transactions, AI adoption is now deep. Ninety-six per cent of dealmakers are using or exploring AI for sourcing and screening. Half have it regularly embedded in due diligence. Forty-three per cent say AI is already making better decisions than humans in some scenarios, and 62% say that relying on human-only decision-making in complex situations is no longer defensible. 

The gains are tangible. Twenty-four per cent of dealmakers say AI has helped them complete transactions they would otherwise have missed. For Raj Bakhru, General Manager at Blueflame AI, the case is direct: "I'd go as far as saying it's negligent to not run your views by LLMs when you're investing hundreds of millions of dollars in a space." 

What is also clear is that AI is not simply absorbing administrative work. It is reshaping how senior practitioners allocate their attention, freeing time for the judgement-intensive work that genuinely requires human input. As Gerrit Beckhaus, Partner at Freshfields, puts it: "It's really crucial to see AI as an enhancer and enabler, to make ourselves as humans better." 

The gap at the top 

There is, however, a consistent pattern in the data. The further up the decision chain, the lower the AI adoption. Thirty-one per cent of dealmakers do not use AI at closing. Twenty-seven per cent do not use it at all for board reporting and governance. 

This is not simple resistance to technology. The data suggests something more deliberate: when it comes to the most consequential decisions, people are holding onto accountability. Only 22% of dealmakers say they would follow an AI recommendation on whether to sign a deal. Forty-five per cent say that decision should remain entirely human. 

Sunil Thakur, Partner at Quadria Capital, captures the reasoning well: "AI can significantly improve analysis and help sharpen decision-making, but at the end of the day, the judgment call is still yours. You have to be very clear about where you trust the AI output and where you rely on your own experience, judgment, and conviction." 

For boards, this is familiar ground. The decisions that reach the board table are, by definition, the ones that could not be resolved earlier in the process. They carry the most complexity, the most contextual nuance, and the deepest organisational consequences. Boards are not positioned to rubber-stamp AI outputs. They are positioned to govern them. 

What boards should require 

The research is clear on what experienced practitioners demand from AI when the stakes are high. Accuracy ranked first (71%), followed by security (70%) and reliability (58%). Speed came fourth. 

That ordering is significant. In EMEA specifically, security marginally overtakes accuracy as the top requirement, reflecting the region's regulatory environment and the sensitivity of the data involved. 

These are not purely technical requirements. They are governance standards. An AI output that cannot be traced to its source, verified against independent data, or audited for its assumptions is not one that a board can responsibly act on. Bakhru is clear about what this demands in practice: "We provide all the ways for a person to fact-check it because it simply must be done. We'll show you how we arrived at that analysis, but humans have to validate it." 

Fifty-eight per cent of dealmakers are already applying human review and validation to AI-generated outputs as a standard step. Forty-four per cent have implemented formal governance frameworks. These are reasonable starting points, and they reflect the same standards of rigour boards apply to any significant input. 

What this means for the future of board decision-making 

By 2030, 23% of dealmakers expect AI to deliver higher-quality decision-making at board level. The top expected benefit at closing is not speed or cost reduction: it is stronger security, governance, and auditability. 

That is a meaningful signal. The expectation is not that AI will accelerate board-level processes in the way it is accelerating due diligence. It is that AI will raise the robustness of the decisions boards ultimately make. Getting there requires the governance infrastructure to be in place now. 

Boards that treat AI governance as an operational detail sitting below their attention will find themselves presiding over processes they cannot adequately scrutinise. Those that engage with it directly, asking what standards apply, which humans are accountable, and how outputs can be verified, will be better positioned to capture the genuine benefits while retaining the accountability the role requires. 

You can download the full report here. The findings are worth reading in full, regardless of whether M&A sits within your board's immediate remit.