Trust, Risk and AI Agents - Rubrik CTO on the New Enterprise Architecture l Dev in the Details Ep 4

Welcome back to Dev in the Details — the show where AI hype gets dismantled and real enterprise strategy takes center stage. In this episode, Rubrik Co-Founder & CTO Arvind Nithrakashyap joins Dev Rishi to expose the real challenges CIOs and CISOs face when adopting AI agents at scale: 👉 Trust 👉 Risk 👉 Governance 👉 Architectural control 👉 And why “just build an agent” is nowhere near enough for the enterprise. Arvind takes us from Rubrik’s early days navigating cyber resilience to today’s AI-native future — including why agents can create 10× value OR 10× damage, what Fortune 500 governance committees are getting wrong, and what an AI-ready enterprise will actually look like in 3 years. This episode is a masterclass for: * CIOs / CISOs / CTOs wrestling with AI risk * Heads of AI, Data & Engineering building agentic workflows * Enterprise leaders struggling to balance speed with compliance * Anyone designing AI governance, risk, or architecture frameworks Topics: ✔ Why AI quality is no longer the top barrier — risk is ✔ The rise of AI governance committees (and why most fail) ✔ How Rubrik is building the enterprise control layer for AI agents ✔ Why observing, auditing & policing agents is unavoidable ✔ The future of AI-native enterprises ✔ Lessons from Rubrik’s origin story that shaped its AI architecture ✔ What CIOs MUST put in place before deploying agents in production 00:00 – Removing toil: What AI should actually do for employees 01:12 – Welcome to Dev in the Details 02:00 – The real bottleneck for AI adoption: Risk, not building 04:15 – Rubrik origin story: Backup, complexity & early insights 07:42 – Enterprise trust: The hardest part of early adoption 10:25 – A law firm, Exchange meltdown & the red-eye that saved the deal13:20 – The importance of founder-level customer obsession 15:05 – Why trust is the keyword for AI in the enterprise 16:30 – What Rubrik expects AI to do for productivity 18:22 – Can AI deliver 2×… or 5× productivity? 20:10 – Understand