AI and DevOps - How to evolve sustainably
Marc Cluet
ABOUT THE SESSION
In this insightful and highly practical session from the Artificial Unintelligence Conference 2025, Marc Cluet challenges the assumption that AI can simply “run operations for us.” Instead, he lays out a clear framework for integrating AI into DevOps in a way that is reliable, accountable, and built for the long term.
Marc explains why sustainable AI adoption depends on the same principles that made DevOps successful in the first place — observability, automation with oversight, clear ownership, reproducibility, and continuous evaluation. He also highlights the risks of rushing models into production without guardrails, and why organizational maturity ultimately matters more than model performance.
Key themes include:
Applying DevOps fundamentals to AI systems
Observability and continuous evaluation for model reliability
Why automation enhances, not replaces, human accountability
Strengthening processes, documentation, and ownership
Balancing speed with sustainability and operational resilience
DevOps, AIOps, MLOps, ArtificialIntelligence, SustainableAI, Automation, Observability, EngineeringLeadership, AUI2025, ArtificialUnintelligence, MarcCluet