UiPath Status

Postmortem: ## Customer Impact

Customers in the Australia, Canada, India, UK, and U.S. regions experienced null values for Input/Output variables in Maestro processes utilizing Human-in-the-Loop (Action Center) tasks between 2026-02-20 02:22 UTC and 2026-02-20 04:23 UTC.

This affected workflows that relied on modifiable input values from Action Center tasks for downstream processing. All other Maestro functionality remained fully operational, and no data loss occurred.

Root Cause

Maestro processes can include Human-in-the-Loop tasks that integrate with Action Center. When a user completes an Action Center task, the input and output values are returned to Maestro for downstream processing.

We deployed a change to address a separate issue where read-only input variables were incorrectly appearing as output variables in Maestro.

What went wrong: Action Center apps support modifiable inputs, which allow users to edit input values during task completion. These modifiable inputs share the same variable names as the original read-only inputs. The deployed change filtered out these modifiable input values along with the read-only inputs, causing them to return as null after task completion.

What should have happened: The change should have distinguished between read-only inputs and modifiable inputs, filtering only the read-only inputs from the outputs.

Detection

The issue was not detected by internal monitoring or alerting systems and was initially reported by customers. It was quickly reproduced internally and confirmed to affect workflows using Action Center tasks with modifiable inputs across multiple regions.

Response

The issue was reported by customers and quickly reproduced internally. The team rapidly identified the root cause and isolated the problem to a specific recent Maestro deployment affecting only workflows with modifiable inputs in Action Center tasks—not all Maestro functionality or all regions were impacted. Once the deployment was identified as the source, the change was promptly reverted to restore the expected behavior.

Follow-Up

To prevent similar incidents in the future, we are taking the following steps:

• Expand test coverage to include additional Action Center workflow configurations, including modifiable inputs.
• Improve coordination and validation between Maestro and Action Center deployments to prevent similar issues.
• Expand automated checks and canary coverage to detect issues earlier, including edge cases involving Action Center integrations.

Resolved: The incident has been resolved.

Monitoring: The fix has now been successfully deployed to all regions. We will continue to monitor system performance to ensure ongoing stability.

Identified: The fix has been deployed to the Community region. We are actively monitoring performance and will provide further updates as rollout continues.

Identified: We've identified the root cause and are currently implementing a fix. We'll post another update as soon as the resolution is live.

Investigating: We are currently investigating an issue affecting Maestro processes that utilize Human-in-the-Loop (Action Center) tasks. After a task is completed, 'Input/Output' variables are incorrectly returning as null. Our engineering team is actively looking into the root cause and we will provide an update as soon as we have more information.

Affected Regions: Australia, Canada, India, UK, and US

Resolved: Issue has been resolved.

Monitoring: We have deployed fix in all regions and seeing improvements..

Identified: We are seeing improvements in performance with the conversational Agents. We are active working on fixing it.

Identified: We have identified the issue. We are working on fixing it.

Investigating: Customers using conversational agent service will experience latency or performance degradation in multiple regions.  We are investigating the issue"

Affected Regions:

• Europe
• US
• Japan
• Australia
• Canada
• Singapore
• India
• UK
• Switzerland

Resolved: Our team has not observed errors in the last 30 minutes, we will continue monitoring.

Monitoring: Our team has not observed errors in the last 30 minutes, we will continue monitoring.

Investigating: We are investigating an outage affecting Document Understanding in East US. Impact: Users may be unable to Classify and/or Extract documents.

Resolved: This incident has been resolved. The issue has been mitigated and services are now operating normally. Thank you for your patience and understanding

Identified: We have confirmed the root cause of the outage impacting Traces UI for Orchestrator in Multiple Regions. Impact: Users may still be unable to view traces.

Users impacted by this issue can view traces from the Agents page.

Identified: We have confirmed the root cause of the outage impacting Traces UI for Orchestrator in Multiple Regions. Impact: Users may still be unable to view traces. Users impacted by this issue can view traces from the Agents page.

Identified: We have confirmed the root cause of the outage impacting Traces UI for Orchestrator in Multiple Regions. Impact: Users may still be unable to view traces.

Identified: We have confirmed the root cause of the outage impacting Traces UI for Orchestrator in Multiple Regions. Impact: Users may still be unable to view traces.

Identified: We have confirmed the root cause of the outage impacting Traces UI for Orchestrator in Multiple Regions. Impact: Users may still be unable to view traces. Next update: Restoration efforts are in progress.

Identified: We are investigating reports that Traces UI for Orchestrator is unavailable in Multiple Regions. Impact: Users may be unable to view Traces UI. Next update: We are actively investigating the issue.

Investigating: We are investigating reports that Traces UI for Orchestrator is unavailable in Multiple Regions. Impact: Users may be unable to view Traces UI. Next update: We are actively investigating the issue.

Resolved: This incident has been resolved. We experienced a service disruption in Canada that affected document classification and extraction. The issue was caused by temporary performance limits in our system. A fix has been implemented, and the service is now fully operational. Thank you for your patience.

Monitoring: A fix has been implemented and we are monitoring the results.

Investigating: We are investigating an outage affecting Document Understanding in Canada. Impact: Users may be unable to Classify and/or Extract documents. Next update: Diagnostic data is being collected to determine next steps.

Investigating: We are investigating an outage affecting Document Understanding in Canada. Impact: Users may be unable to Classify and/or Extract documents. Next update: Diagnostic data is being collected to determine next steps.

Resolved: DU Metering errors have not recurred in the past 3 hours. We are continuing to monitor for any further occurrences before closing this incident.

Monitoring: DU Metering errors have not recurred in the past 3 hours. We are continuing to monitor for any further occurrences before closing this incident.

Monitoring: DU Metering errors have not recurred in the past 3 hours. We are continuing to monitor for any further occurrences before closing this incident.

Monitoring: DU Metering errors have not recurred in the past hour. We are continuing to monitor for any further occurrences before closing this incident.

Monitoring: DU Metering errors have not recurred in the past hour. We are continuing to monitor for any further occurrences before closing this incident.

Investigating: We are continuing to investigate an outage affecting DU Metering for Document Understanding in Canada. Impact: Users may be unable to access DU Metering. Next update: Diagnostic data is being collected to determine next steps.

Investigating: We are investigating an outage affecting DU Metering for Document Understanding in Canada. Impact: Users may be unable to access DU Metering. Next update: Diagnostic data is being collected to determine next steps.

Resolved: Our cloud service provider has confirmed that the issue on their end has been fully resolved. Our service has fully recovered and is stable.

Monitoring: We are currently not observing new errors related to this issue, and service performance appears to be stabilizing. However, the underlying root cause has not yet been fully confirmed by our cloud service provider. We continue to work closely with them to validate the cause and ensure a complete and sustained resolution.

We will share further updates as more information becomes available.

Identified: Customers using Anthropic Sonnet 3.5v1 may experience issues with the Agents service. The issue has been identified with our cloud service provider, and we are actively working with them to resolve it as quickly as possible.

Resolved: This incident is now resolved. The status sync issue is fixed and the backlog of stuck tasks has been completely cleared.

Monitoring: The remediation effort has successfully completed, and previously stuck workflows have been updated to reflect their correct status. New Agent tasks continue to process normally. We are now monitoring the system to ensure complete stability across all regions before marking this incident as resolved.

Identified: The cleanup effort is nearing completion, with the vast majority of affected tasks now reflecting their correct status. We are continuing to process the small remaining backlog of "InProgress" workflows and will update you shortly.

Identified: We are still actively working to correct the statuses of the remaining affected Agent tasks. The cleanup process is ongoing, and we will keep you informed as we make further progress. Thank you for patience.

Identified: We have deployed a mitigation for this issue. Newly initiated Agent tasks via Maestro are now correctly reporting their completion statuses. However, our engineering team is still actively working on correcting the status of existing workflows that were impacted and remain stuck in "InProgress." We will provide another update once this effort is complete.

Identified: We have identified an issue affecting status reporting for Agent tasks started via Maestro. These tasks may appear stuck in "InProgress" even after completing successfully. This issue is impacting all regions. We are currently working on a resolution and will provide an update shortly.