🎥 Recorded live at the MLOps World | GenAI Summit 2025 — Austin, TX (October 8, 2025) Session Title: Don’t Page the Planet: Trust-Weighted Ops Decisions Speaker: Eric Reese, Senior Manager, Site Reliability Engineering, Best Buy Talk Track: AI Agents for Model Validation and Deployments Abstract: Enterprises don’t need more dashboards—they need AI-driven deciders that act safely. In this talk, Eric Reese presents how Best Buy built an agentic validation layer that sits between ML predictions and operational responses, making automated decisions trustworthy, auditable, and adaptable. The system uses exponentially decaying trust scores with configurable half-lives, dynamically adapts thresholds by time, scope, and diversity, and classifies incidents before routing them through policy-driven workflows. When confidence drops, “gray-zone” cases are escalated intelligently instead of triggering unnecessary alerts. Eric breaks down the Predict → Categorize → Weight → Accumulate → Decide → Notify pipeline—a production-tested framework for transforming noisy ML outputs into governed, high-confidence actions. The session also covers trust accumulation algorithms, multi-agent validation patterns, and observability tools for auditing every decision safely. What you’ll learn: • How to build trust-weighted operational agents that act with confidence and control • How to combine validation, thresholds, and routing to eliminate alert fatigue • How to design adaptive, context-aware incident classification • The architecture behind Best Buy’s agentic decision layer for production ML systems • Practical lessons for context-driven and safe AI operations.