If you were to ask executives how well they believe their managers handle hard conversations, 74% are confident that they can handle them alone with ease. Yet when managers reflect on their own experience, 42% said they knew exactly what to say but couldn’t land it. Another 26% couldn’t predict how the other person would respond. And 20% could plan for everything except the moment the conversation went sideways.
CEOs and managers, describing the same moment, are reaching entirely opposite conclusions about how well those tough conversations go. Leadership confidence in their managers quietly redirects budget and attention elsewhere, away from the managers still learning how to handle a conversation that doesn’t go according to plan.
Data from PI’s AI and People Leadership Report shows just how consequential that can be. When leaders overestimate how prepared their managers are, they risk missing the support managers need.
Are CEOs solving the right problem with AI?
There are several layers to what managers struggle with most when preparing for a difficult conversation, including delivering criticism constructively (23%), anticipating employee reactions (20%) and remaining objective (14%).
Each of those struggles comes down to managers not knowing how the specific person across from them thinks or reacts. When managers feel uncertain, they often ask whatever AI tool is on hand how to phrase the feedback, predict the reaction or steer the conversation back on track.
Generic AI advice can’t fill in the blank, because it’s built to answer everyone the same way. Believing managers are ready doesn’t make them ready, and neither does a tool that hands out the same advice regardless of who’s in the room.
An overwhelming majority of executives (94%) intend to invest in AI tools for managers, even as they underestimate how much support those managers actually need to close on any of the struggles above. Without a clear problem to solve, the tools they fund are likely to sharpen managers who already deliver these conversations well, while leaving managers who struggle with them unsupported. Tools aimed at the wrong problem stall adoption, and the investment doesn’t pay off.
How can HR choose people-first AI tools during an AI rollout?
The tools worth funding do one thing generic AI can’t: Give managers more human context in their AI. That means accounting for how a specific person behaves, anticipating how they’re likely to respond, and helping a manager adjust when a conversation takes an unexpected turn.
The judgment still belongs to the manager. What changes is the information they’re working with.
Leaders already see the value of this kind of support. Ninety-one percent of leaders agree that AI tools can give managers good advice if they have sufficient context on each employee. That same confidence carries straight into how much they’re willing to invest — in tools that put personalized, science-backed coaching in every manager’s hands before feedback conversations and team conflicts.
That’s where PI comes in. Backed with behavioral science, PI’s AI provides the human context managers need to handle difficult conversations and adapt their approach based on the person in front of them.
What would better support look like for managers?
Better support gives managers information about the specific person they’re preparing to face, such as how that person tends to communicate, what motivates them and how they’re likely to behave under pressure. A generic prompt can’t produce any of that. It only pulls patterns from everyone else who’s ever asked a similar question.
A manager’s own experience with that person is valuable on its own. A behavioral profile adds a consistent read on how that person operates, built from validated data instead of a handful of past interactions. It doesn’t replace what the manager already knows. It gives them a starting point when the history isn’t there yet, and a second reference point when it is.
HR can get ahead of the AI rollout by setting requirements before the budget moves, starting with asking vendors what behavioral data shapes their advice, and prioritizing vendors whose guidance changes based on the person in the room. Executives are already willing to fund tools for managers. HR can direct the funding toward vendors who meet the bar, and away from tools built to sound the same for everyone.
Managers already feel pressure to use AI without a clear standard for what “good” looks like. Behavioral data gives that standard shape, and lets HR fund technology that protects culture instead of straining it further.