No incidents in progress. Uptime figures cover the trailing 90 days and are measured from outside our own network.
| Service | Status | 90-day uptime |
|---|---|---|
| API | Operational | 99.98% |
| Dashboard | Operational | 99.99% |
| Support Triage | Operational | 99.97% |
| Sales Qualifier | Operational | 99.96% |
| Research Analyst | Operational | 99.91% |
| Document Extractor | Operational | 99.95% |
| Data Cleaner | Operational | 99.98% |
| Meeting Summariser | Operational | 99.94% |
| Content Writer | Operational | 99.93% |
| Workflow Runner | Operational | 99.90% |
| Billing | Operational | 100% |
Runs against document sets over 50 files took up to four times longer than usual for about 90 minutes, caused by a context-assembly regression. No runs failed and nothing was overcharged, since cost follows tokens rather than duration.
A queue worker deadlock delayed webhook deliveries by up to 40 minutes. Runs themselves were unaffected and every delayed delivery was eventually made. Retry logic now has a watchdog.
A session store migration caused intermittent login failures for 25 minutes. The API was unaffected throughout. No credential data was exposed.