AI is accelerating software delivery—but it’s also changing the risk profile of how we ship software. In this episode of Ship Happens, Ivar Conradi Østhus, CTO and creator of Unleash, joins Per Krogslund to explore how Feature Ops helps teams ship faster and safer through controlled rollouts, feature flags, and real production feedback.
Software delivery is getting faster—but also more unpredictable. In this episode of Ship Happens, we sit down with Ivar Conradi Østhus, CTO of Unleash, to unpack how AI is reshaping the way teams build, ship, and operate software. At the center of the discussion is “Feature Ops,” a philosophy that combines DevOps practices, feature flags, and real-world feedback loops to enable safer, faster delivery cycles. Ivar explains why production is not the end of the pipeline—but the beginning of learning. We explore how teams use DORA metrics to measure delivery performance, and why AI adoption—while increasing developer velocity—can introduce new risks around deployment stability and system reliability. To counterbalance this, Ivar breaks down the importance of runtime control, observability, and automated feature management as essential safeguards in modern engineering systems. The conversation also dives into how AI is already being used to automate operational tasks like feature flag cleanup across distributed systems. Finally, we discuss how enterprise development tools are evolving—moving closer to developer workflows like pull requests, CI/CD, and runtime decision-making—and why the future is likely a best-of-breed ecosystem rather than a single unified platform.
We discuss: What “Feature Ops” actually means in practice, How DevOps and DORA metrics measure real delivery performance, Why AI increases both velocity and risk in software delivery, The role of feature flags in safe production releases, Runtime control, observability, and production learning loops, AI-driven automation in DevOps workflows, Enterprise tooling integration into CI/CD pipelines, Platform engineering vs best-of-breed ecosystems, The future of AI in developer productivity and delivery safety. Whether you're building engineering systems, leading platform teams, or navigating AI adoption, this episode explores what it really takes to ship safely in the age of AI.
Key Topics Discussed
Episode Timestamps
(00:00) AI Threat Framing
(00:52) Welcome and Guest Intro
(02:00) What Is FeatureOps
(04:03) Experiments and Fast Feedback
(06:06) DevOps Metrics and DORA
(09:55) Funding and AI Risk
(14:04) Runtime Control and Rollouts
(17:14) Testing Renaissance and Trust
(20:46) Accountability and AI Ethics
(24:35) AI Automation and Flag Cleanup
(28:41) Agents and Smarter Software
(29:36) Natural Language Integrations
(31:07) MCP Versus CLI Debate
(31:56) Enterprise Security Concerns
(33:51) Standards and AI Translators
(37:16) SOAP Makes a Comeback
(38:23) SaaS Valuations and AI Fear
(41:28) What Great Dev Tools Do
(44:54) Platforms Versus Best of Breed
(47:48) Platform Teams and Enterprise Data
(50:18) Measuring AI Productivity ROI
(54:29) Pricing Reality and Model Choice
(56:44) Wrap Up and Sponsor Thanks
Guest Bio
Ivar Conradi Østhus is the CTO and creator of Unleash, one of the leading open-source feature management platforms used by engineering teams worldwide. He has spent his career building developer tooling and scaling software delivery systems, with a focus on helping teams ship safely through feature flags, controlled rollouts, and runtime decision-making. Ivar is a leading voice behind “Feature Ops,” a modern approach to DevOps that emphasizes production-first learning, fast iteration, and strong operational guardrails in complex engineering environments.
Links & Resources
Per Krogslund’s LinkedIn
Ivar Conradi Østhus’s LinkedIn
Learn more about Unleash
Learn more about Docker