Many sponsors now expect AI to show up in the value-creation plan. The ambition is reasonable. The results are often uneven. In my experience, the reasons have less to do with technology than with how the work is introduced into a company.
Why rollouts stall. They start with the tool, not the economics. An initiative that begins with a platform decision, rather than a specific result the business needs to move, tends to produce activity without impact. Teams adopt features; the P&L does not notice.
They compete with everything else. Portfolio companies are usually already carrying a demanding plan: new systems, new leaders, add-ons, reporting requirements. AI arrives as one more priority for the same few people. Without explicit capacity, it slips.
Nobody owns the outcome. When AI sits between the sponsor, the CEO, IT, and outside vendors, accountability blurs. Pilots start, but nobody is responsible for turning them into changed operations.
The organization is not ready. Disconnected systems, undocumented processes, and unreliable data do not stop AI work entirely, but they shape what is realistic. Plans that ignore this tend to promise too much in the first year.
Management is not aligned. If the CEO and department heads see AI as something being done to them rather than with them, adoption stays shallow. A sponsor mandate is not the same as operating buy-in.
What works instead. Anchor every initiative to a measurable economic goal. Margin, cycle time, revenue per employee, customer retention: pick the result first, then decide whether AI is the right lever.
Sequence the work to the company's capacity. One or two well-chosen initiatives that actually land are worth more than a broad program that stalls. Early, visible results build the credibility for the next step.
Put one accountable operator on it. Someone needs to own both the economic target and the practical change inside the business. That person has to be able to work with the CEO and department heads, not around them.
Read the organization honestly. Assess leadership, processes, systems, and data before committing to timelines. The plan should match what the business can actually absorb.
Scope to the situation. Not every mandate is portfolio-wide. Sometimes the right engagement is a single platform company, a specific workflow, or a post-close priority. The discipline is the same at any scale.
As a former COO, I have seen that the plans that work are the ones matched to the organization's ability to execute. AI does not change that principle. If anything, it makes it more important.