Prioritization as Code: An AI-Supported Framework for Platform Engineering

Platform Engineering teams face a tough question: how do you decide what to work on when your users are internal, your resources are limited, and stakeholders care about different things like security, cost, speed, and reliability? In the Platform Engineering organization at SiriusXM, we treated this like a product problem. Supporting about 700 internal developers, we needed a consistent, repeatable way to prioritize our work based on real user needs and measurable impact. Over two years, we built and refined a framework that has guided 200+ platform investments. It connects our roadmap to company strategy using user research and data—not opinions—and translates inputs like developer pain points, reliability risks, and executive cost concerns into a shared scoring model that makes tradeoffs clear and transparent.

Along the way, we ran into a common issue: teams were making decisions based on different assumptions about users and data, which led to roadmap conflicts. To fix this, we used AI to synthesize feedback and signals from multiple sources and surface the assumptions behind our decisions, with platform leaders providing context and governance. We then codified those validated assumptions into what we call "Prioritization as Code"—a living artifact that documents how we prioritize and evolves as we learn. In this session, we'll share the full framework, our impact rubric, how we use AI to validate assumptions, and practical lessons from 200+ decisions so you can adapt the approach to your own team and build alignment when priorities compete.