Monitoring and service levels (hosted service)

• Availability is checked from outside every 30 seconds, from three regions; two minutes without an answer raise an alert.
• Service levels are evaluated every five minutes: the rate of server errors, response times of reads, workflow moves and agent tools, failed agent calls, document renditions, and how fresh the backups are. A breach and its recovery are mailed to the operators (since 2026.11.0).
• Capacity, measured (2026-10-30): one instance of today's size served about 25 agent calls a second within those levels, which is roughly 80 to 150 agents at work. No organization's data appeared in another's answers, in about 14,000 calls under load.
• Incidents. A written procedure covers response and notifying customers: a customer whose personal data is affected is told within 24 hours of confirmation. There is one responder, stated plainly in that procedure. The procedure has not yet been exercised.