Knowing what to do and actually doing it under pressure are two very different things. Marc Bernstein built Balto around that gap. In this episode of _At Work, Under Pressure_, Jessi Beyer sits down with Marc Bernstein, CEO of Balto, to talk about AI, contact centers, human connection, real-time coaching, de-escalation, and what happens when someone knows the right skill but their brain fails to retrieve it in the moment. Marc shares how his own experience in a contact center inspired Balto: he could receive great training, understand exactly what he was supposed to do, and then finish a live customer conversation realizing he had forgotten to do it. That distinction between knowledge and real-time performance eventually became central to the company’s approach. Jessi and Marc discuss: • Why skills ultimately matter more than scripts, while scripts can still provide useful examples of what good performance sounds like • How real-time AI support can reduce the cognitive load on contact center employees and free them to focus on human connection • Why empathy becomes harder when employees are simultaneously worrying about whether they are “doing empathy correctly” • How AI could help replace the traditional churn-and-burn contact center model with higher-value human work • Why Balto deliberately tries to keep AI out of the “human lane” • What frustrated customers actually need in order to de-escalate • How AI may expand human creativity and capability rather than simply replace expertise Marc also shares a particularly relevant example from one of Balto’s crisis-center customers: a new employee encountered a caller expressing suicidal ideation, Balto detected what was happening and alerted the supervisor in real time, and the supervisor was able to leave a meeting and come support the employee through the interaction. This episode is for contact center leaders, customer experience teams, sales leaders, AI executives, learning and development professionals, and anyone interested in what happens when technology is designed to support human performance rather than replace it. Because training someone is not the same thing as preparing them to perform. And the most useful technology may be the technology that gives humans enough cognitive space to do the things only humans can do.