Knowledge v2 is down
Timeline · 2 updates
- investigating Oct 18, 2026, 01:40 PM UTC
Knowledge v2 went down.
- resolved Oct 19, 2026, 07:17 PM UTC
Knowledge v2 recovered.
Aitronos had 15 outages in the last 2 years totaling 129h 34m of downtime, averaging 0.6 incidents per month.
There were 15 Aitronos outages since September 18, 2026 totaling 129h 34m of downtime. Each is summarised below. Incident details, duration, and resolution information.
Knowledge v2 went down.
Knowledge v2 recovered.
Smart Documents is degraded.
Knowledge v2 recovered.
Smart Documents is degraded.
Smart Documents recovered.
Chat errors is degraded.
Smart Documents recovered.
Chat service is degraded.
Chat service recovered.
Chat replies were slow or failing for some customers using one model. Chat kept working for everyone else.
Chat replies are back to normal for the affected customers.
File sync on one Mac briefly returned errors. Chat was not affected.
File sync is back to normal.
Document preview images were delayed because a background rendering queue fell behind. Everything else worked normally.
Preview rendering has caught up.
Heavy database load caused about 9 percent of requests to fail for roughly seven minutes, including some page loads and sign-in refreshes. Chat replies kept working and most requests succeeded.
All requests are succeeding again.
Some chat replies took longer than a minute to start. Replies were still delivered.
No further failed downloads were seen.
Chat replies failed repeatedly in a single conversation. Chat worked normally for everyone else.
Chat is back to normal.
Knowledge v3 went down.
Knowledge v1 recovered.
Knowledge v2 had errors processing files for a small number of knowledge stores. Most uploads and knowledge searches worked normally.
Knowledge v3 is back to normal.
Platform went down.
Knowledge v2 recovered.
Requests to api.aitronos.com timed out from all probe regions between 19:02:53 and 19:06:23 UTC. The cause was a cold start: outside business hours the API runs on a single warm instance, and when the platform recycled that instance there was no warm sibling to absorb traffic. Service recovered on its own once the replacement instance finished starting. No data was lost. Monitoring has since been adjusted so a recycle of this kind is confirmed over a longer window before it is treated as an outage.
Resolved. api.aitronos.com returned to normal response times at 19:06:23 UTC and has been healthy since.