I've spent my career between software and data engineering. Two sides often work is silos and think the other is doing it wrong.

When I started working with coding agents, I kept running into same issues as everyone else: stale specs nobody owned, missing context the agent couldn't discover, no trace of why it made a decision. They were new, yet felt familiar...

It's because they are data governance problems applied to the context of software engineering. Ownership, discovery, contracts, lineage, quality. The vocabulary is different but the issues are similar... Ans so are solutions.