I don't think model drift is the only issue. In production, I've seen more failures caused by changes around the model, such as APIs, retrieval quality, data pipelines, permissions, and business logic. Even with a stable model, the surrounding system continues to evolve. That's why observability and continuous evaluation at the application layer remain essential.
I completely agree. That observation is actually what led us to build StateGuard. We expected model failures to dominate, but after reproducing a few dozen real-world issues, most of them turned out to be contract mismatches between APIs, providers, schemas, tool outputs, and application logic.
We're trying to understand which of those failures developers would actually want repaired automatically versus simply surfaced through observability.
Genuine question: should model providers eventually solve this at the source, or is defending against drift always the application layer’s job?