August 2026

August 2026 release notes for Orchestrator in Automation Cloud.

August 18, 2026

Run Agents and Functions on local robots in preview

Note:

Correction, August 20, 2026: Maestro jobs are not currently supported on local robots.

Agent, Function, and API jobs can now run on unattended robots that you host yourself, instead of only on Cloud - Serverless.

Machine templates gain a new Agents and Functions runtimes section, where you configure Agent and Function capacity independently of your existing RPA runtimes. Running these jobs on local robots is consumption-based, similar to Cloud - Serverless licensing, but doesn't factor in machine size.

For details on configuration, routing, and licensing, see Agents and Functions on Local Robots.

Pin the Execution settings panel open when starting jobs

In the Start job dialog, a new pin toggle in the Execution settings panel header lets you set your preferred default expansion state. When pinned, the panel opens expanded every time you start a job, so you no longer need to re-expand it on each run. The preference is saved per user and follows you across folders.

Configurable job recording duration

A new Maximum recording duration (minutes) field in the Job recording section lets you set the maximum video recording duration, pre-filled with your tenant's default and validated against your tenant's limit. Non-entitled cloud tenants remain capped at 3 minutes. Tenants on the Enterprise plan (Unified Pricing) or the Advanced plan (Flex) can set a duration up to 15 minutes. For more information, see Deploying processes.

Job stop reason visible in Job State History

The Job state history timeline now displays the reason a job was stopped, surfacing information Orchestrator already recorded internally but did not previously show. For more information, see Viewing job details.

Support for multiple package feeds

Orchestrator now supports configuring multiple NuGet feeds, as well as npm registries and PyPI indexes, instead of a single feed per category. For more information, see Deployment tab.

Queue item failures by reason on the per-queue monitoring dashboard

The per-queue monitoring dashboard now includes a Queue item failures by reason widget, sourced from Insights real-time monitoring data. Failures are aggregated by failure reason and exception type (Application or Business), with an inline drill-down to the individual failed queue items for a given reason. This widget replaces the classic Error Feed widget where available. For more information, see Monitoring queues.

Early validation when the personal workspace automation limit is reached

When you try to publish a solution to your personal workspace (Publish for me) and your entitlement has already been reached, the publish screen now shows an error instead of failing silently. The error links you directly to Deployments, where you can delete an existing automation to free up your limit. For more information, see Personal workspaces.

Focus tree toggle in the job graph

In the job details side panel, the graph displayed at the top of the Details tab now includes a Focus tree toggle when there are more than 10 nodes. Turning it on highlights the direct path from the selected node up to the root node, making it easier to trace a specific job through large graphs. For more information, see Viewing job details.

August 12, 2026

Automation Cloud Robots - Serverless static IPs for the United States — delayed update (GxP) region

You can now add static IPs to the allowlist for Automation Cloud Robots - Serverless in the United States — delayed update (GxP) region, instead of opening your network to all external IPs. To use static IPs, you must explicitly enable this option in your serverless robot template.

After enabling them, you can add the corresponding IPs to your allowlist. For the list of static IPs, see Outbound IP ranges.

Automation Cloud Robots - Serverless region-level redundancy in Japan

Automation Cloud Robots - Serverless in the Japan region now have redundant deployments across two geographically isolated regions, Japan East and Japan West. Previously, they ran in Japan East only. Both regions are located in Japan, so data residency is unchanged. As a result of this change, the Japan region has additional outbound IP ranges.

If you allowlist static IPs for serverless robots in Japan, you must add the new ranges to your allowlist. For the full list, see Outbound IP ranges. For details on how UiPath approaches redundancy, see High availability and disaster recovery strategy.

August 3, 2026

New storage bucket provider: Google Cloud Storage

Add Google Cloud Storage (GCS) as a new storage bucket provider in Orchestrator. Configure authentication using a service account key. For more information, see Managing storage buckets.

Azure Storage bucket authentication using service-principal connections

Authenticate Azure Storage buckets using Integration Services service-principal connections instead of storage-account access keys. This enables customers' security teams to enforce Entra-based identity and access management policies. For configuration details, see Managing storage buckets.

Tenant-level Jobs view: add column parity with folder-level view

The cross-folder Jobs view now includes all columns available in folder-level Jobs views, including Started/Ended (relative and absolute), Creation Time, Source, Priority, Machine, Hostname, User interaction, and Healing Agent status. All added columns default to hidden; enable them via the Columns dropdown. See Jobs monitoring for details.

Side panels now overlay primary content

Sidepanels now render on top of primary content instead of pushing it aside. Panels are resizable via a drag handle, with user-chosen width persisting per session. No backdrop is displayed, so interactions with the underlying grid or list continue while the panel is open.

Debug jobs now display generic Process label

Debug jobs now consistently display the [Studio Debugging] Process label regardless of whether they are Desktop or Web session types. This ensures a uniform debugging experience across job types.

OpenAPI Public profile is now the default

The Orchestrator default API specification is now compliant with the OpenAPI 3.0 specification. The previous specification (V20, based on Swagger 2.0) is deprecated as of this release and will be removed in September 2026. For more information and the full deprecation timeline, see Read me.

Queue trigger: batch processing disabled for Maestro processes

When configuring a queue trigger for a Maestro process, the Minimum number of items to trigger the first job and Another job is triggered after each fields are now fixed to 1 and cannot be changed. See Queue triggers for details.

Automation Cloud Robots: clearer machine size label

In the Additional Settings step of process configuration, the machine size fields for cross-platform processes are renamed for clarity: Cloud Robot - Serverless is now Cloud serverless machine size. The Job Priority selector no longer displays a required-field asterisk that incorrectly marked the field as mandatory. For more information, see Managing processes.