Reported impact
Business outcome
Tastien gave its nationwide restaurant estate a repeatable way to turn noisy monitoring into actionable incidents. The new workflow compresses alert floods before they reach responders, cuts invalid alerts, shortens recovery time, and keeps structured root-cause history so each resolved incident improves the next diagnosis. Instead of relying on individual operator memory, the company can now reuse AI-assisted runbooks and feedback data across business lines.
- Productivity Improvement
- Risk Reduction
- Scalability
Solution
Tastien turned restaurant-operations alerting into an AI-assisted incident workflow with Alibaba Cloud Cloud Monitor 2.0 and STAROps. ARMS, CMS 1.0, SLS, Prometheus, certificate alerts, and custom business alarms are all ingested into CMS 2.0's unified event center, which replaces siloed alert channels with one normalized entry point. On top of that, Tastien added a low-code workflow layer that parses source-specific payloads, standardizes titles, severities, and resources, deduplicates repeated incidents, and routes alerts to the right Feishu groups and on-call schedules. STAROps then uses the alert context, log and monitoring data, and a service-specific root-cause library to produce AI diagnostics, let operators question incidents directly from the alert card, and run digital employees for recurring inspection and RCA work. Human feedback on actual root causes and AI accuracy is fed back into the same loop so the system improves instead of stopping at first-response automation.
Data flow
Alerts from CMS, ARMS, SLS, Prometheus, certificate, and custom business sources enter CMS 2.0, where low-code workflows standardize fields, deduplicate incidents, and route live cards into Feishu. STAROps then reads the alert context together with logs, metrics, and historical root-cause entries, returns AI diagnosis to operators, and stores human feedback back into the same incident history for future reuse.
