The Constraints Will Move. You Have to Learn Your Way Into Them

Building software got much cheaper and faster with AI agents. Everything upstream and downstream of the software build portion has not. Instead of solely focusing on coding speed gains, we’ve spent the past year working on different bottlenecks in the value stream: faster decisions on what to build, writing better requirements, improved validation, and cutting out things we used to do just because that’s how we did it.
I lead engineering at a logistics technology consultancy running client engagements with increasingly smaller teams and shorter iterations. I’m convinced that evolving technology requires evolving practices. We continue to use Scrum, but I’ll explain why we dropped some of the surrounding exercises and practices that no longer seem valuable to us.
This is an experience-based talk on my findings rather than a framework discussion. I'll cover a number of things that didn't go well early on: AI-written backlog items, pull requests touching a hundred files, no review agent in the loop. I'll also touch on a prediction I got wrong about a role I assumed AI would eliminate first. It didn't.
Topics will include:

• Where the delivery bottlenecks transitioned to after coding got faster
• What we got wrong when we first put AI into our SDLC
• Why continuous delivery practices like working in small batches, trunk-based development, and fast rollbacks matter more than ever
• What we kept and cut from our agile practices
• Spec-driven development, and whether it fits teams like ours

About your speaker:
Kevin Jenkins is VP of Engineering at Metafora, a logistics technology consultancy building custom software for the transportation and logistics industry. He leads a fully remote engineering practice delivering client engagements across freight rating, carrier integrations, and transportation management systems.
He's been building software for 25 years across e-commerce startups, healthcare, advertising, and the music industry, and has led a couple of waterfall-to-agile transformations along the way.