Most conversations about AI stop at capability: what can the model do? The more important question for anyone running a business is availability: will it still be doing it at three in the morning, on a holiday, under load, when its inputs have quietly changed?
Reliability is a property of the system, not the model
The model is one component among many. Whether the whole thing stays up depends on everything around it: timeouts and retries, graceful degradation when a dependency is slow, sensible fallbacks when the model is uncertain, and hard limits that stop a bad day from becoming an outage. None of this is AI work. All of it is what makes AI usable in production.
Observability for systems that are allowed to be wrong
Traditional monitoring assumes a clear line between correct and broken. AI systems live in between: technically up, quietly producing worse answers than yesterday. That demands a different kind of observability — tracking output quality, latency, cost, and drift, not just uptime — so that a slow degradation is caught as an alert rather than a complaint.
The pager is part of the design
An operable system is designed with its worst day in mind. That means runbooks for the failures you can predict, alerts that page a human before customers notice, the ability to roll back a model as easily as a deployment, and clear ownership of who responds when it fires. Deciding all of this after launch is how a promising system becomes one nobody wants to maintain.
We build with the pager in mind from the start, because operating the system is not an afterthought to the engineering. It is the engineering. That is what it means to build software that survives contact with reality.