Scaling AI: Optimization, Observability & Evaluation

מור אלדר, Senior AI Specialist, AWS | ארז ויינשטיין, Solutions Architect, Startups, AWS | נעמה דמתי, VP Research, CX Division, NICE

AWS Summit Tel Aviv 2026 · Building AI and Agentic Applications · 38 min · 17 pages

As AI usage grows, keeping it reliable, cost-effective, and explainable gets harder: costs climb faster than traffic, dashboards stay green while answers go wrong, and no one can explain what an agent did overnight. This session shows how observability, evaluation, and optimization work together to scale AI on Amazon Bedrock with confidence — from optimizing cost and performance, to gaining visibility into what your models and agents actually do, to confirming that quality holds as you evolve. Hear from NICE on the observability they built into production, and the lessons they learned along the way.

Generated automatically from the recording's transcript. It's meant to replace re-watching, not the recording itself — verify against the video at the timestamp before quoting formally. Slides belong to the speakers and organisers. Produced with recording-to-pdf.