AI tooling and adoption have been exciting. AI tooling and adoption have been terrifying. How can engineering organizations thread the needle?

For the past few decades, teams have been told there is a zero-sum tradeoff between innovating ("move fast and break things") and shipping software correctly and safely. This makes no sense in practice, and it's exactly why agents are making unauthorized changes in production right now. It is a pernicious ethos to bring to modern software development, and it should be abandoned immediately.

With the explosion of agentic engineering, agents are shipping at the same velocity as engineers, if not greater, but without the judgment or awareness engineers are expected to have. It does not have to be this way, nor should it be.

As organizations enter an era of unprecedented velocity, it's well past time to reimagine this approach as safe velocity. This session will explore the foundations and patterns — context, guardrails, and skills — that make this possible in practice. By putting these in place from the outset, engineers are empowered to experiment and ship with confidence. Safe velocity isn't a constraint on innovation; it's what makes innovation sustainable.