Session Abstract

Your organization has Agents everywhere - chatting with customers, updating records, summarizing meetings - and everyone’s proud… until one of them confidently rewrites your HR policy. This session is for teams that know “we’ll handle governance later” really means “we’ll be in damage control forever.” We’ll move past static documents and checkbox compliance to show how to design, document, and enforce governance for Copilot Studio Agents using the Cloud Adoption Framework (CAF) and enterprise best practices. This is about making governance real, not ornamental. You’ll learn how to: - Define governance objectives aligned with CAF disciplines - policy, identity, cost, operations, and security. - Turn written policies into living controls embedded in environments, roles, and ALM pipelines. - Apply Responsible AI principles - data boundaries, transparency, and content moderation that actually work. - Automate guardrails with environment policies, DLP rules, and deployment gates. - Build a Center of Excellence that enforces consistency without suffocating innovation. Because the real question isn’t “Who gave the Agent admin rights?” - it’s “Who thought governance was optional?”

Katerina Chernevskaya

🏅 Microsoft MVP, MCT, and Power Platform Solution Architect with 20+ years in tech. 🏗️ Builds enterprise solutions with Power Platform, Copilot Studio, Microsoft Foundry, and Azure AI. 🤖 Particularly interested in the moment an AI agent stops answering questions and starts doing things - and the architecture needed to keep that under control. 🎤 Speaks internationally about AI, governance, and the messy gap between a great demo and a production system. 🧪 Mentors new tech talent and believes the best way to learn something is to build it, break it, and share what happened. 📜 Collects certifications too: Power Platform, Azure AI, Prosci, and PMI.