Wolf&Grayson EditorialWhile 88% of organizations are already using AI, only 10% of boards are considered AI-ready. That gap is a governance failure, one with a measurable price tag. Boards with proper AI oversight outperform those without by a factor of five. Alvaro Celis, a former Microsoft executive who managed billion-dollar profit and loss (P&Ls) and now serves on corporate boards, has spent his career at the intersection of capital allocation, technology strategy, and enterprise value creation. His read on what is happening in boardrooms right now is direct. “The most dangerous directors are not the ones with deep technical experience,” Celis says. “They are the ones without it, because they participate in consequential decisions without the ability to meaningfully influence the outcome.”
Approving AI Is Not the Same as Understanding What You Just Approved
When a board approves a capital project, the framework is established. Milestones exist, compliance checkpoints are built in, and the norm is understood. When a board approves an AI initiative, it is not approving an investment. It is the approval of a transfer of capabilities, judgment, and decision-making authority. Most boards have not registered that distinction, and the gap between what they approved and what they understand is where liability accumulates.
The second-order risks compound the problem. Decision-making accountability that dissolves the moment a system, rather than a person, makes the call. In Europe, where AI is classified as a product rather than a project, boards that approve AI initiatives without demanding auditability and traceability are not just making poor governance decisions; they are walking into legal exposure. “You’re delegating authority without control,” Celis says. “And that is a fundamentally different risk than anything most boards have managed before.”
Automation Is Not the Strategy. Reimagination Is
The boards treating AI as a cost-reduction exercise or a technology line item are not thinking about the right problem. The organizations that will define competitive advantage in the next decade are not the ones that simply apply AI to existing processes. They are the ones that use AI to reinvent their business models entirely. The multiplier logic that Celis observed in building Microsoft’s partner ecosystem applies directly here. For every dollar earned by Amazon Web Services (AWS) on AI, its partners generate seven. That network effect – the interdependency, stickiness, and value created across an ecosystem rather than within a single company – is the strategic opportunity most boards are not yet discussing.
“A platform is an economic layer where other companies make more money working with you than you make selling your own product,” Celis says, citing the foundational lesson from Bill Gates. The question every board should be asking is not how to use AI to cut costs – it is how to use AI to create the kind of ecosystem value that compounds.
Agentic AI Requires Governance That Most Boards Have Not Built
In 2026, 75% of businesses plan to implement some level of agentic AI. Given the current state of board-level AI governance, where 74% of boards do not discuss AI at every meeting and two-thirds of directors have limited AI knowledge, the resulting risk landscape is significant. Celis is direct about the consequence; boards that apply traditional capital allocation models to the agentic AI transition will create serious liability for their organizations.
The governance framework he advocates has three non-negotiables.
- Auditability, the ability to explain and trace every AI decision, a legal requirement in multiple jurisdictions.
- Kill-switch capability, the ability to immediately shut down automated systems and revert to backup processes when something goes wrong.
- Human accountability that cannot be delegated even when execution is automated.
“You can delegate execution,” Celis says. “You cannot delegate accountability.” A board that lacks the first two and misunderstands the third is not governing AI. It is approving it and hoping for the best. The next decade will produce two categories of companies: those whose boards treated AI as a line item, and those whose boards used it to reinvent how they create value. “You don’t need the best AI model to win,” Celis says. “But you absolutely need an AI-empowered, disciplined board. That will be a foundational requirement for success.”
Follow Alvaro Celis on LinkedIn for more insights on board governance, AI strategy, and enterprise value creation in the AI era.
