Building AI You Can Stand Behind: Governance, Security, and Trust in Production (w/ The AI Collective)

Building AI You Can Stand Behind: Governance, Security, and Trust in Production You’re building with AI. You’ve got a prototype. Maybe your team is already shipping models, deploying agents, or integrating AI into real customer workflows. But once AI moves from demo to production, the questions change. How do you govern systems that are constantly evolving? How do you move fast without cutting the wrong corners? What does “responsible AI” actually mean when product, security, legal, compliance, and business teams all define risk differently? Join AI Collective Seattle for a candid fireside conversation on what it really takes to build trustworthy AI systems in the real world. Expect a practical conversation on what separates responsible AI in theory from responsible AI in production. We’ll cover the decisions that never make it into policy documents, the tradeoffs teams face when shipping quickly, the frameworks that matter, the ones that can slow teams down, and the new security challenges created by agentic and frontier AI systems. Speakers Tristan Ingold | AI Governance Advisor Tristan works as an AI Governance Advisor, with a background across security governance, risk management, compliance, and product risk. His work focuses on the governance challenges emerging at the intersection of frontier AI, enterprise risk, and product development. He is also a LinkedIn Learning instructor, with courses on AI governance, the EU AI Act, cybersecurity compliance, and GRC reaching more than 75,000 learners. Tristan brings the perspective of someone working inside large-scale technology organizations, where AI governance is not an abstract policy conversation, but a real-time product and risk discipline. Joe Braidwood | GLACIS Technologies Joe is the CEO and Co-founder of GLACIS Technologies, a Seattle-based AI security startup backed by AI2 Incubator. GLACIS is building infrastructure for AI runtime assurance, including tamper-proof records and cryptographic evidence that help companies prove their AI safety controls actually ran as intended. GeekWire described GLACIS as creating “flight recorder” style infrastructure for enterprise AI systems. Before GLACIS, Joe co-founded Yara AI and previously helped build Scener, a co-watching platform that scaled to millions of users. His work now sits at the center of one of the most important questions in AI: how can companies prove that AI systems behaved safely, not just claim that they did? Ken Johnston | AIGovOps Foundation & Envorso Ken is a founder of the AiGovOps Foundation and Vice President of Data, Analytics, and AI at Envorso. With more than 30 years of experience across Microsoft, Ford, and Autonomic.ai, he has led work spanning cloud platforms, data science, software quality, telemetry, and MLOps. His current work focuses on turning AI governance from policy into infrastructure: embedding automated controls, observability, and operational accountability directly into how AI systems are built and deployed. Moderated by: Bhola Chhetri | Orena Discussion Themes Why AI Governance Matters Now Most AI governance conversations happen at the policy or compliance level. This conversation will focus on what governance looks like at the product level, while systems are actively being built, deployed, and scaled. Operationalizing Governance We’ll explore what “good” looks like for AI teams in practice. How do teams move from checklists to operating models? What should founders and engineers take from frameworks like the EU AI Act, NIST AI RMF, and ISO 42001, and what can they safely ignore early on? Security, Risk, and Edge Cases AI systems introduce new kinds of failure modes. We’ll discuss how AI security differs from traditional security, what agentic AI changes, and how teams should think about incidents, audit-ability, data boundaries, and accountability. The Future of Responsible AI As AI systems become more autonomous and more deeply embedded in business workflows, governance and secur