De-risking Deployments in an AI World
summary
Claims that LLMs must be treated as untrusted and that safe automation requires hardened CI/CD pipelines: enforce zero trust, mandate unit test coverage, performance and chaos testing, and shift policy enforcement into the pipeline. Cites an infrastructure team doing millions of deployments daily as proof automation scales. Describes a policy engine built on Open Policy Agent (Rego) that warns or blocks deployments and using AI to generate Rego/policies to remove tedious work — outcome: faster, safer rollouts when guardrails exist.
AI-written summary · about
chapters (10)
- 0:00De-risking AI deployments
- 1:05Beyond DevOps automation
- 3:13Securing generative AI
- 5:59AI governance and policy
- 9:11Data governance risks
- 13:45AI in continuous delivery
- 16:32Removing mundane tasks
- 18:38AI policy engines
- 20:22AI risk and maturity
- 23:52AI adoption and risk
related talks
45mDeploy First, Secure Never: AI Between Productivity and Security Chaos
30mSecuring AI Workloads: Building Zero-Trust Architecture for LLM Applications
12mThe AI Trust Triad for DevOps
22mRethinking Risk, Governance, and Compliance Applying Automation
29mBuilding Trust in AI Agents
46mLeveraging AI and Automation in DevOps
36mHow to Safely Deploy AI Agents that Write and Manage your IaC
20mBuilding Deterministic Infrastructure for Non Deterministic AI Agents