The Board Approved An AI Strategy: Do They Know What It Requires?
Nacho De Marco is Chairman and Co-Founder of BairesDev, a leading software development partner, and co-founder of VC firm BDev Ventures.
gettyAs CEO of a company built around tech talent and embedded alongside delivery teams across more than 500 clients, the conversation surrounding AI skills is one I follow closely. It almost always focuses on the people doing the work, how engineers, managers and teams adapt as tools reshape their roles. That conversation matters, but there’s a version of it that should happen at the level where it matters first.
Where AI is stalling, the root cause is rarely the model, the tooling or the team. It’s a decision that didn’t get made, or a question that didn’t get asked, several levels above where the work lives. The knowledge required to make those calls well is neither purely technical nor purely organizational. It’s both, and most leadership teams haven’t had to develop it until now.
AI literacy in its conventional sense, understanding how the technology works, is paramount. At the leadership level, the literacy gap I keep seeing is different. It begins with recognizing that every strategic decision about AI is also a decision about the organization required to support it. This article is about what happens when that understanding is missing, what it costs in practice and what the current landscape demands of leaders navigating AI with assumptions built for a past era.
In 2025, 83% of S&P 500 companies disclosed AI as a material risk, while fewer than 3% of board directors disclosed AI expertise. Organizations are recognizing the stakes. The boards governing them are not yet equipped to match them.
That challenge plays out at three levels simultaneously.
Organizations are disclosing AI as a material risk and treating it as a strategic priority, but the ownership question beneath that momentum is largely unanswered. In a recent survey we conducted with over 1,300 developers across 61 countries, only 10% of developers said accountability for AI outcomes sits with senior leadership. Over half said it sits with them personally.
I see this across client organizations constantly. No one sets out to leave accountability undefined. It just never gets named, and by the time anyone notices, the team closest to the work has already absorbed it.
In a recent study we conducted that surveyed over 500 technology leaders, 79% said they feel pressure to overstate AI progress, and nearly half said that pressure originates from the C-suite or board. The result is a feedback loop where leadership sets aggressive targets, teams present a rosier picture than reality supports and leadership then makes the next round of decisions on inflated data. What I find most damaging about this cycle is that by the time anyone has accurate information, the decisions it should have informed are already made.
Our study linked above found that deadlines and delivery pressure are the top barriers to validating AI output. Leaders who compress timelines without adjusting quality expectations are making a quality decision by default rather than by design.
This is the least visible and the most consequential layer. One fellow at our company frames this challenge as an inverted pyramid. He found that roughly 70% of what AI success requires is people and process change, 20% is tooling and 10% is the models themselves. Budget and attention flow in the opposite direction.
Much of the institutional knowledge that systems depend on lives in people’s heads, not in documentation or data pipelines. AI exposes that problem quickly. A model operating against incomplete context produces confident output that sails through review because the reviewer is missing the same context the system is. In my experience, this is where organizations most often mistake an institutional knowledge problem for a technology problem.
Our recent study mentioned above found that 88% of technology decision-makers saw at least one active AI initiative significantly disrupted by shifting executive priorities in the previous year. What appears downstream as an execution problem often begins well above the delivery team.
The three gaps above each trace to a leadership decision that either got made too late or didn’t get made at all. Addressing them is where AI literacy at the leadership level becomes operational.
Before approving an AI initiative, assess the data foundations, available capacity and process dependencies it will run on at production scale. If those preconditions aren’t in place, fund them first or adjust the scope to what the organization can actually support.
Define what validation requires (testing standards, review gates, quality thresholds, etc.) as part of the project approval, not as something the delivery team negotiates against the deadline. When leadership specifies what rigor looks like upfront, teams can stop trading quality for speed.
Name who owns AI outcomes before the work starts and recognize that ownership is shared. Business leadership owns the value question, whether the initiative is worth pursuing and how success gets measured. Technical leadership owns the readiness question, whether the organization’s knowledge, context and infrastructure are captured well enough for the system to deliver on that value. Neither side can carry the other’s responsibility.
Execution quality matters at every level. Teams must remain responsible for what they ship, but the conditions that determine whether that work succeeds are a leadership responsibility.
Boards do not need to become AI engineers, but they do need to understand what the organization must change before an AI strategy can work. Approving the strategy without addressing those conditions leaves delivery teams to absorb decisions leadership did not make.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
