Back
AI Program Manager – Customer Experience
Full-time
At Saviynt, customer support runs on autonomous AI agents that triage, route, and resolve cases without human intervention. The next frontier is measurable: higher deflection, faster closure, and agents that learn from every interaction.
As the Business Analyst for Customer Support (AI), you are the analytical engine that tells us exactly where our agents are winning, where they're failing, and what it will take to push autonomous resolution rates higher. You track agent performance at a granular level, identify patterns in cases that escape to humans, and feed those findings directly into the tuning cycles that make our AI sharper over time. This is a high-ownership, high-visibility role where your insights directly shape how agents are trained and how productivity gains are measured.
What You Will Be Doing
Own performance analytics for the AI agent fleet, tracking deflection rates, autonomous closure rates, escalation triggers, confidence score distributions, and CSAT by resolution type.
Dig into cases that escape AI resolution, identifying intent clusters, knowledge gaps, and routing failures, and translating findings into improvements for agent training and prompt refinement.
Define and govern the KPI framework for AI productivity, including human-time-saved per case and agent-assisted vs. fully autonomous closure ratios.
Build and maintain real-time dashboards surfacing AI agent health, case closure velocity, and deflection trends.
Run structured QA and eval cycles on AI agent outputs, reviewing response accuracy, tone calibration, and resolution completeness.
Prepare concise, evidence-based briefings for leadership on AI agent performance and productivity gains.
What You Bring
3-6 years of experience in analytics or business intelligence, with time spent analysing AI, ML, or LLM-powered systems in production.
Hands-on experience with AI agent evaluation: designing rubrics, running QA cycles, and defining what good resolution means in a support context.
Sharp analytical instincts, able to distinguish improving deflection driven by better agents from frustrated users giving up.
Clear communication, turning complex AI performance analysis into crisp, actionable recommendations.
Familiarity with prompt or context engineering, enough to have an informed opinion on why an agent responded as it did.
