Patial outage impacting for computer vision in US region

Update timeline

1. investigating May 13, 2026, 03:06 PM UTC

   We are investigating reports of an outage impacting for computer vision in US region. Next update: Our teams are working to understand the cause and scope and will share updates as available.

2. identified May 13, 2026, 04:04 PM UTC

   The issue has been identified, and our teams are actively working toward applying a fix

3. monitoring May 13, 2026, 04:16 PM UTC

   The issue has been mitigated and resolved. Services are now operating normally. We will continue to monitor the environment to ensure stability.

4. resolved May 13, 2026, 04:31 PM UTC

   The issue has been mitigated and resolved. Services are now operating normally. We will continue to monitor the environment to ensure stability.

5. postmortem Jun 03, 2026, 06:27 AM UTC

   ## Customer impact Between approximately May 13, 2026 at 14:50 pm UTC and May 13, 2026 at 16:16 pm UTC, a subset of customers using Computer Vision in the U.S. region experienced elevated request failures. The affected requests returned server errors and timeouts, which could cause automations that depend on Computer Vision to fail or require a retry. ## Root cause The incident was caused by a failure in an internal routing component \(the Location Service\) used to direct traffic across UiPath Cloud services. Following a routine restart, new instances of this service were unable to complete startup successfully — the internal configuration used for routing traffic failed to load up in the allocated time and new instances failed to start. Most UiPath services were largely shielded from this failure because they route traffic through a layer that caches Location Service data, allowing them to continue functioning normally even when the Location Service is degraded. Computer Vision, however, uses a different routing path that does not benefit from this cache and must reach the Location Service directly for each request. As a result, Computer Vision in the U.S. region was disproportionately impacted, returning errors and timeouts to customers, while other services largely continued to operate. ## Detection The issue was detected through an automated high-severity incident alert for the Computer Vision service on May 13, 2026 at 14:58 pm UTC, approximately eight minutes after the estimated start of customer impact. ## Response An incident bridge was opened for investigation. The status page was updated at 15:06 UTC to reflect that Computer Vision in the U.S. region was under investigation. At approximately 16:00 UTC, Computer Vision error rates increased further when the remaining healthy instance was replaced during the restart cycle. The response team increased the service compute resource allocation, allowing new instances to complete startup successfully. Within minutes, they passed health checks and began serving traffic, and by ~16:07 UTC error rates had returned close to baseline. The incident was fully mitigated at 16:16 UTC and marked resolved at 16:31 UTC ## Follow-up 1. Ensure enough capacity is allocated to the service for it to handle spinning up new instances. 2. Improve our service startup check's timeout settings to ensure the service has enough time allocated for loading its configuration and starting properly.