We Track Changes and Decisions. We Don’t Track Intent - and AI Makes It Worse.

We’ve built systems that perfectly track what changed (Git) and why decisions were made (ADRs), but we still fail to capture intent-what the system was actually supposed to achieve. This gap forces engineers to reconstruct context from logs, wasting time and leading to subtle failures. As AI accelerates code generation and system changes, this problem gets worse: more code, less preserved reasoning. To build reliable systems, we need to treat intent as a first-class artifact, not something inferred after the fact.