How do you tell users your AI agent is down?
Serious question. If you're running an agent in production (customer support bot, coding assistant, data pipeline), what happens when it breaks at 3 AM? Traditional status pages track HTTP endpoints. They don't understand model providers, agent…