Cyara’s CEO outlines AI agent oversight as a workforce model

Cyara’s platform supports over 350 million customer journeys annually across 140 countries, focusing on testing, monitoring, and validating AI-driven customer interactions. Kumar stresses that enterprises must treat agent autonomy as a progression, requiring evidence of competence before granting more responsibility. He warns that traditional software testing metrics fail to capture AI agent failures, which often result in customer dissatisfaction rather than system errors. Instead, enterprises must evaluate outcomes—not just responses—and treat production failures as mandatory release gates to prevent repeated mistakes.