March 2026 Release Notes

Insight Explorer is being upgraded with a new option for the process attributes setting  that utilizes an AI agent to discover process inefficiencies that may impact the behavior of your KPIs. Identifying these inefficiencies will improve both the causality and quality of the insights being generated while also allowing for more complex process patterns in your insights, such as activity sequences or ping pong behaviors. Those process inefficiency attributes replace the basic process attributes.

To activate the process inefficiency attributes agent in your Insight Explorer, click the Advanced Settings button in the upper right corner. In the Advanced settings window, select the “Enhance with AI-detected inefficiencies” option under “Attribute recommendations” and then click the Update button.

Once this Insight Explorer is configured and deployed, the process inefficiencies will be generated and displayed alongside your standard attribute recommendations.

For more information, see Configuring an Insight Explorer.

Process Copilots can now generate structured responses by utilizing a fixed JSON schema that can be customized in the AI Instructions of your Process Copilot configuration. Generating structured responses instead of the text strings provides a more predictable and repeatable output for your integrations with Action Flows, Orchestration Engine, or external APIs. Using this new configuration, you can develop a schema that maps specific fields directly to your Action Flow, Orchestration Engine, or external system without manual parsing. These structured responses prevent the need for manually coding JSON responses while also avoiding the potential processing errors caused by the string responses in your downstream analytics.

These new schemas are created on the AI Instructions section of your Process Copilot configuration by enabling the Structured data output toggle. You can then manually enter your schema or click the icon in the right corner to ask the AI assistant to generate the schema Schema based on your description. Once you have finished configuring this Process Copilot, you can test the schema and verify how your responses are displayed using the Integrations tab.

For more information, see AI Instructions.

You can now define and execute authenticated API calls to external web services directly within your Process Orchestration steps.

This new capability allows you to collect relevant data from external services and seamlessly inject it back into your orchestration workflow, making your automated processes more flexible and integrated with systems outside of Celonis.

For more information, see Adding cloud functions to Process Orchestration steps.

You can now use Process Copilot capabilities directly in your Process Orchestration steps, bringing conversation AI into your automated end-to-end workflows.

This integration allows you to use Process Copilot's ability to converse with your Celonis data within any orchestration step. You can analyze data by asking natural language questions, build graphs, search for insights, or view process flows, making your automated business processes more dynamic and data-driven.

For more information, see Adding Process Copilot to Process Orchestration steps.

We have enhanced the Send Email action in Views. In addition to using your personal email address, you can now send messages using the Celonis service address (noreply@celonismail.com).

For more information, see Adding Actions to Studio components.

We've made some performance improvements in Views. The content in your View will no longer reload when you enter and exit View edit mode, giving you a more seamless user experience.

In the Extractor Builder, you can now use the new Filtering Method configuration to define specific filter rules for endpoints, similar to the existing pagination setup. By selecting specific methods based on your source API, you now have more granular control over your data queries, such as using individual query parameters for standard requests or the request body for POST endpoints.

For more information on configuring filtering methods for endpoints, see Configuring endpoints.

With this Public Preview release, we've made some changes to how you interact with Knowledge Models, making them easier and more intuitive to use.

We’ve updated the Knowledge Model asset editor. By replacing the current left-hand vertical navigation with a new horizontal tabbed navigation, we've improved screen real estate use.

Within the Knowledge Model asset editor, you can also now show/hide table columns including short display name, internal note, unit and format. We've also introduced advanced filtering at column level:

You can also now access the new PQL editor from directly within a Knowledge Model using the three dot button:

For more information about working with Knowledge Models, see Knowledge Models.

And for more information about our public preview releases, see Feature release types.

The metrics available for the Multi-Object Process Explorer are being expanded with the new Event Count tab which allows you to see how often an event occurs for a specific object. This new tab is available by clicking on any event in your Process Explorer graph to view the event node details.

This tab includes an expandable section for each object and includes the count metrics for this event, including the average number of times it occurs, the maximum number of times it occurred, and the total number of occurrences for this object. Each section also includes a graph showing the frequency with which every number of occurrences happens for this object, such as how often this event occurred four times or how often it occurred twice. This information is similar to the metrics shown on the “Event count” tab in the Event Explorer metrics.

Along with this change, the average times per object metric is being removed from the Occurrence tab in the node details.

For more information, see Event node details.

Process Copilots are being updated to include a new option that allows you to switch between sources (External API and Celonis Asset) in order to test both the internal and external integrations of your Process Copilots from the chat interface. This update is designed to give you full control over how you test and validate your Process Copilots during the development process and allows you to simulate different execution surfaces in order to ensure that your configuration delivers accurate reasoning.

On the Integrations panel, you can select to use the External API option to test conversations triggered from third-party applications such as Microsoft Teams and then switch to Celonis Assets to verify task executions triggered from your Action Flows or Orchestration Engine. The results of the tests from either integration will be logged and can be viewed from the Monitoring page.

As part our recent updates to versioning and deploying packages, we added a badge to the user interface which displayed the version and deployment status of packages and other Studio assets. We've now discovered an issue with this status badge so have removed it, meaning the status badge will no longer display.

For more information about versioning and deploying assets, see Versioning and deploying packages.

We've added an object and event filters in the event log builder. These filters allow you to select specific events or objects to include or exclude in your event log. By reducing the amount of data materialized in the log, this results in faster performance and more efficient analysis of process mining data.

This improvement is especially useful for large datasets, where unnecessary events or objects can slow down analysis. You now have more control over the granularity of the data included in your logs, enabling cleaner and more focused insights.

To learn more about creating event logs in the Knowledge Model, see: Event logs (object-centric).

We've made some changes to improve the consistency of filters across Views in the Celonis Platform.

Previously, some Views filters were applied and saved automatically as soon as values were selected, with the filtered output updated each time an individual value was selected or removed. For other Views filters, the filter was applied and saved only when confirmed via an Apply button.

We've now aligned all Views filters so filters are only applied and saved when the Apply button is selected. A Cancel button also allows the filter selection to be easily discarded.

For more information about Studio filters, see Filtering Studio Views.

We have expanded the PyCelonis DataModel with a dedicated Augmentation API. You can now programmatically manage the entire lifecycle of an augmented attribute directly from your Jupyter notebooks:

Previously, injecting machine learning predictions or custom Python calculations back into the data model required routing the data through a complete data integration pipeline and triggering a data load. With this release, you can now close the loop, writing results straight from a Pandas DataFrame in MLWB into your operational data model in real-time.

Ensure your PyCelonis library is updated to the latest version to access the new Augmentation API. For detailed code examples and syntax, please refer to the PyCelonis documentation and the Augmented Attributes user guide.

You can now track your team's Celonis Compute Units (CCUs) usage from your data pool overview screen. Celonis Compute Units are a standardized metric used to quantify the compute capacity required on the Celonis Platform when using the new ETL Engine.

When monitoring and tracking your CCU usage, the following details apply:

To learn how to track your CCU usage, see: Tracking CCU usage.

And to learn more about these metrics, see our terms and conditions: Celonis - Terms and Conditions - Services.

With this public preview release, you can now configure an Instance Explorer component for your object-centric data in your Studio Views. The component lets you explore individual object instances, like Orders, Invoices, or Flights, and their events within the digital twin. You can see each object’s lifecycle and relationships without creating custom queries or tables.

The key features include:

The Instance Explorer component can be viewed using a graph or list view, with an example of the graph view displayed here:

To learn more about configuring and using the Instance Explorer component, see: Instance Explorer.

And to learn more about your public and private product releases, see: Feature release types.

The new Celonis Assistant is an AI-enabled support assistant that guides you in the use of your Celonis products in real-time. The Celonis Assistant can offer best practices, suggest products, provide assistance with troubleshooting or guide you to build custom solutions on the Celonis Platform.

The Celonis Assistant can be accessed from the main menu on the Celonis Platform and opens a chat window allowing you to ask questions. Using this chat functionality, the Celonis Assistant can offer help with product specifications, best practices, product updates, troubleshooting, use cases, success stories, suggested solutions and more. As you explore the various processes available for you to implement, the Celonis Assistant can provide suggestions about KPIs to track or specific use cases to analyze, and share implementation success stories.

To use the Celonis Assistant, you will need to enable AI Settings and Azure OpenAI models for your team. For more information, see AI Settings.

With this public preview release, you can now assign custom alias names to objects and events within your perspectives. Using aliases allows you to simplify the naming conventions and shorten the more complicated technical table names when creating PQL statements. Once an alias has been assigned, you can use both the original name and the alias when creating your queries.

To enable the use of aliases when viewing your perspectives, click Options - Enable display names:

To learn more about creating and using perspectives in Objects and Events, see: Perspectives.

And to learn more about our public preview releases, see: Private/Public Preview features.

The architecture of your Process Copilots has been upgraded to unlock deeper thinking capabilities and provide more reliable data handling. This overhaul will remove technical bottlenecks and allow your Process Copilots to deliver more accurate responses and process even more complex workflows with ease.

Dimension switching enables faster, more flexible self-service analysis, allowing users to explore multiple data perspectives within a single chart for quicker insights and a better experience.

Users can now dynamically adjust the data in bar, column, line, and area charts by enabling dimension switching.

Previously, users couldn't easily switch between visualizing data on a weekly and a monthly basis.

With dimension switching, users can now change the data perspective displayed using a simple dropdown in the chart.

In this example, the user can select whether to display the Sum data by Ship Mode, State or City.

To find out more, see Switching dimensions in charts.

You can now migrate your full Analysis assets and settings to Studio Views with the migration wizard, enabling you to create dashboards and reports using our enhanced View editing experience.

While Celonis continues to maintain Analysis (by fixing bugs and ensuring performance remains strong), no further feature development will take place. We therefore recommend you migrate your Analysis assets to Studio Views.

For more information, see Full analysis to Views migration.

You can now migrate bookmarked filters from Analyses to Views.

Previously, bookmarked filters used in Analyses had to be manually recreated from scratch in Views, resulting in additional work for users.

Admin and analyst users with edit permissions can now migrate bookmarked filters as required during the Analysis to View migration and select whether to migrate all bookmarked filters or shared bookmarked filters only.

Once migrated, the bookmarked filter type and owner are respected in the View and shared with the same users in the View as in the original Analysis.

For more information, see Migrating bookmarked filters to Views.

You can now import translations from a Data Model to a Studio package.

Translations in an Analysis use Data Model name mappings. In Views, translations are saved in the package settings. To migrate translations, the name mappings can now be imported to the package settings.

When migrating an Analysis to a View, the translation syntax is automatically updated during the migration and uses the translation keys from the package, with no changes required in the Analysis or View. Users can then select their preferred language from the View Translations menu.

For more information about migrating Analyses, see Migrating Analysis to Views and for information specific to this functionality, see Migrating translations from an Analysis to a View.