2026-09-22 –, Main Track All times in US/Mountain
DevOps has made pipelines fast, automated, and highly reliable. We deploy more frequently than ever before. Yet systems still fail in production, not because we cannot deliver quickly, but because we often build the wrong behavior.
Spec-first approaches, user stories, and even modern testing practices validate assumptions too late. By the time gaps, edge cases, and unintended interactions are discovered, teams are already deep into rework, defect cycles, and operational risk.
This talk introduces Simulation-Driven Software Engineering (SDSE) as the next step beyond spec-first thinking. Instead of relying on static requirements, SDSE enables teams to create executable behavioral simulations before code exists. These simulations model real-world conditions, including failures, timing, and complex interactions across systems.
Drawing on real-world experience, this session shows how simulation can expose hidden assumptions early, align stakeholders on actual system behavior, and dramatically reduce defects and rework.
The goal is not to slow down DevOps, but to enhance it.
Because a fast pipeline only delivers value when it delivers the right system.
Your Pipeline Is Fast… But Is Your System Right? From Spec-First to Simulation-Driven Software Engineering
If you have spent time in the DevOps community, you already know how to make software delivery fast. We have CI/CD pipelines, infrastructure as code, automated testing, observability, and increasingly, AI helping us write and ship code even faster.
So why do we still see production issues that feel… avoidable?
This talk starts with a simple observation: speed is not the same as correctness. We have optimized the system for flow, but we often validate behavior too late. Specs, user stories, and acceptance tests still leave room for interpretation. By the time reality shows up, we are already committed to an implementation.
What I will explore with you is a shift I have seen work in practice, long before it had a name, and now more relevant than ever in an AI-driven world: Simulation-Driven Software Engineering (SDSE).
Instead of asking, “Did we build it right?” after code exists, SDSE asks, “Do we agree on how this system behaves?” before we write it.
We do that by creating executable simulations of system behavior. Not static diagrams. Not long documents. Living models that can be run, tested, and challenged by the whole team. These simulations include real-world complexity such as failures, latency, concurrency, and unexpected interactions across services.
In this session, I will walk through:
Why spec-first and test-after approaches still lead to rework and defects
How simulation exposes hidden assumptions across Dev, Ops, Security, and business stakeholders
What a simulation-first workflow looks like in a modern DevOps environment
Where this fits alongside CI/CD, testing, and platform engineering rather than replacing them
This is not about adding process. It is about moving validation earlier, where it is far less expensive and far more effective.
I will also connect this to something many of us are seeing right now: AI is accelerating development dramatically. That is a good thing. But it also increases the risk of building the wrong thing faster. Simulation becomes a way to bring discipline and shared understanding back into the system without slowing teams down.
If you have ever thought, “We built exactly what was asked, and it still failed,” this talk is for you.
The goal is simple:
Keep your pipelines fast. Just make sure they are delivering the right system.
Marc Hornbeek, a.k.a., DevOps-the-Gray, is Senior DevOps Advisor for The DevOps Institute, CEO and Principal Consultant at Engineering DevOps.com, Founder of Intelligent-Institute.com, Author of books Engineering DevOps, Continuous Testing, Quality, Security, and Feedback, Intelligent Continuous Security, Engineering Respect and Trust, and Simulation-Driven Software Engineering. Advisor, ambassador, author and instructor for PeopleCert - The DevOps Institute. Blogger on DevOps, CloudNativeNow, SecurityBoulevard, and Platform Engineering and LinkedIn websites. Marc is a globally recognized strategic consultant who applies disciplined engineering practices, holistically, for Continuous Testing, DevOps, DevSecOps and SRE digital transformations. Marc has led more than 95 transformations for enterprises, manufacturers, service providers and government institutions. He is an IEEE Outstanding Engineer, and an IEEE Life member. His education includes engineering and executive business degrees and numerous DevOps, DevSecOps and SRE certifications.