Managers Turn to AI to Rehearse Firing and Pay Talks

New AI coaching tools are helping managers prepare for some of the hardest conversations in the workplace, telling an employee their performance is falling short, denying an expected raise, or eliminating their position, letting managers rehearse dialogue, get feedback on tone and wording, and model different employee reactions before the real conversation happens, according to CNBC's reporting on the trend.

A July survey from The Predictive Index, a company that offers a text-based chatbot called Obi specifically designed to assist with role-playing and coaching managers through difficult conversations, surveyed 399 managers and 208 CEOs and business leaders, revealing a genuinely notable perception gap. While 74% of CEOs and business leaders say their managers are "very confident" handling tough people conversations independently, only 42% of managers themselves say they actually know what they want to say but struggle with exactly how to phrase it. Just 20% of surveyed managers say they prefer preparing for a difficult conversation entirely on their own, without a framework, coaching guide, or HR input.

Emily DeJeu, a professor of business communication at Carnegie Mellon University's Tepper School of Business, was direct about why treating these conversations as something to handle spontaneously is a mistake. "Spontaneity is not a goal of a high-stakes work conversation," DeJeu said, according to CNBC's reporting. "You want to know the executive has really thoughtfully prepared for an important conversation." The survey identified specific manager blind spots worth understanding directly, including delivering criticism constructively, anticipating an employee's emotional reaction, and staying genuinely objective throughout the exchange.

Key Findings From The Predictive Index Survey

The tools on the market vary genuinely widely in their specific capabilities, spanning free and paid options, some text-only and others voice-enabled, with some incorporating specific employee data such as personality profiles to help tailor the AI's coaching more precisely, according to CNBC's reporting. Managers can direct the AI to play the role of an emotional or angry employee reacting to feedback, and the system can separately evaluate a planned message against specified criteria, flagging legally risky language, excessive wordiness, repetition, or filler words. Xactly Chief People Officer Megan Barbier described the practical value directly: the tools can provide individualized scenarios and recommendations in real time, tailored to a manager's specific situation.

The most consequential detail in this trend isn't the coaching benefit itself, it's a real, documented privacy risk connected directly to how managers are actually using these tools. According to CIO Dive's separate reporting on the same underlying survey, 72% of managers say public AI tools help with these conversations, but 44% admit to entering employee names or specific performance details directly into those public tools, even though only 45% of organizations surveyed have a formal, written policy governing AI use in performance management at all. That gap, high adoption paired with low formal governance, connects directly to the broader AI privacy and data handling concerns we've tracked closely across multiple industries this year.

Specialists cited in CNBC's reporting flagged a specific, technical limitation worth understanding directly: chatbots may be inclined to simply agree with the user rather than offer genuinely critical feedback, meaning managers need to actively and explicitly ask for candid, critical assessment rather than assuming the AI's validation reflects an objectively well-prepared approach.

This trend is worth understanding for any business currently developing or updating its own AI usage policies, since the specific gap identified here, high manager adoption of public AI tools for sensitive personnel conversations alongside minimal formal governance, represents a genuine, immediate compliance and data privacy exposure most organizations haven't yet addressed directly.

For businesses in HR technology or management training specifically, this data offers a clear signal that purpose-built, privacy-conscious AI coaching tools represent a genuine market opportunity distinct from managers defaulting to general-purpose public chatbots with employee data, a distinction worth building directly into any internal AI policy guidance.

How are managers actually using AI for difficult workplace conversations?
Managers are using AI chatbots to rehearse dialogue for conversations about performance issues, denied raises, or layoffs, letting the AI role-play as an employee reacting to feedback and evaluating planned messages for tone, clarity, and legally risky language.

Are managers putting employee data into these AI tools safely?
Not necessarily. A survey by The Predictive Index found 44% of managers admit entering employee names or performance details into public AI platforms, even though only 45% of surveyed organizations have a formal written policy governing AI use in performance management.

Why do experts recommend preparing for these conversations rather than being spontaneous?
Carnegie Mellon business communication professor Emily DeJeu says spontaneity isn't the goal in high-stakes conversations, and thoughtful preparation helps managers deliver criticism constructively, anticipate emotional reactions, and stay objective throughout the exchange.

Managers are increasingly using AI coaching tools to rehearse difficult workplace conversations about performance, pay, and layoffs, with a July survey from The Predictive Index finding 72% of managers say public AI tools genuinely help with these conversations. The same survey revealed a real privacy risk: 44% of managers admit entering employee names or performance details into public AI platforms, even though only 45% of organizations have a formal, written policy governing this kind of AI use. Experts recommend managers actively seek critical feedback from these tools rather than assuming AI validation reflects genuinely thorough preparation, given chatbots' documented tendency toward agreeable rather than genuinely candid responses.