AI duty engineer for production troubleshooting
What 3AM wants to solve is the issue of being woken up by the police at three o'clock in the morning. It is an AI duty engineer deployed in the enterprise intranet: it is connected to Prometheus, Grafana, Datadog, PagerDuty, Kubernetes and various databases. When an alarm comes, it goes to the online system to collect evidence. It first proves the root cause, and then provides a repair plan, which is approved by the person before execution.
It targets enterprises such as banks with strong supervision and old systems that cannot be changed, so its selling points are all about security: local deployment, no inbound ports, manual approval of every operation, and audit logs to prevent tampering. There is currently a 90-day free trial, with a maximum of 10 instances installed. It is a bit heavy to use for online services of small and medium-sized teams. Financial and telecommunications systems with high operation and maintenance pressure can try it.