Capabilities

Platform updates

Functions

Ontology

Updated language models are now available in TypeScript functions repositories. These updates provide better consistency between model APIs, making it easier to interchange underlying models. Model capabilities have also been enhanced, with improved support for vision and streaming. Users can now migrate from existing language model imports to access these new features. Note that existing language model imports are being deprecated in favor of the updated versions.

## Updated language models now available in TypeScript functions repositories

[Updated language models](../functions/language-models.md#language-models-within-functions) are now available in TypeScript functions repositories. These updates provide better consistency between model APIs, making it easier to interchange underlying models. Model capabilities have also been enhanced, with improved support for vision and streaming. 

We highly recommend updating your functions repositories with the new models to ensure you stay up to date with the latest AIP features. Review the updated documentation for [language models in functions](../functions/language-models.md#language-models-within-functions) to learn how to update your repository.

![Viewing model capabilities when importing updated language models.](./media/2025-07-16-172747-functions-image-pn.png)

_Viewing model capabilities when importing updated language models._

### Share your feedback

Share your feedback about functions by contacting our Palantir Support teams, or let us know in our [Developer Community ↗](https://community.palantir.com/) using the [`functions` tag ↗](https://community.palantir.com/tag/functions).

Updated language models now available in TypeScript functions repositories

Pipeline Builder

Data Integration

Pipeline Builder now automatically generates descriptive names for all node types in your graph, not just transform nodes. This makes it easier to quickly create meaningful names for joins, splits, and other node types based on their function or content.

Support generation of names in preview header for all column types

Upgrade Assistant

Administration

The Upgrade Assistant overview page now displays the ratio of pending-to-total actions for each upgrade, rather than completed-to-total actions as previously shown. This change makes it easier to identify upgrades requiring action.

Upgrade Assistant displays pending-to-total action ratio

Upgrade Assistant now categorizes upgrades into six types to help you better prioritize and plan your upgrades. The new types are: Admin action, Model migration, Platform change, Remediation, Security, and Version update. These upgrade types are visible on the Upgrade Assistant overview page.

New upgrade types in Upgrade Assistant

AIP Evals

AIP Evals now supports static values for evaluation suite columns in object set-backed test cases. Previously, you could only associate object references, object properties, or directly linked object types with evaluation suite columns. With this enhancement, you can provide a single static value that applies to all test cases in an object set, giving you more flexibility when configuring your evaluation suites.

Support static values for object set parameters in AIP Evals

AIP Logic

AIP Logic now displays cURL link icons for functions and actions, providing quick access to API call syntax. This allows you to easily integrate AIP Logic functions into external applications or scripts. The links are disabled for ontology edit functions.

Add cURL to call the Logic Function using Foundry APIs

Workshop

App Building

The Vega Chart widget now correctly detects selections made across overlapping layers within a chart and accurately updates the output variables of each selected layer. 

Accurate selection detection across overlapping layers in Vega Chart widget

Code Repositories

The VS Code Extension now automatically loads your previously saved input filters. When you open a file with input filtering options, your saved filter configurations will be preloaded, eliminating the need to manually reapply them each time.

VS Code Extension now displays saved filters

VS Code Extension

Developer Toolchain

Spark Profiles are now supported in the Preview Engine, enabling you to quickly preview Python transforms with custom-defined Spark configuration values. This ensures consistent behavior across preview and build functionalities when working with Python transforms. When previewing transforms in VS Code, relevant Spark profile configurations are automatically applied, particularly those needed for runtime behavior and backward compatibility during breaking changes. Note that Spark profiles cannot adjust preview resource allocations - to modify resources for preview, use the settings available on the Code Workspaces settings page.

Spark Profiles support in Preview Engine for Python transforms

You can now access AIP Evals directly from Agent Studio with a new integrated sidebar. This integration allows you to create and manage evaluation suites for your AIP agents without leaving Agent Studio. The sidebar enables you to create evaluation suites, toggle between different suites, edit test cases and evaluators, run evaluations, and view results—all within the Agent Studio interface. This integration is similar to the existing AIP Evals integration with AIP Logic, making the evaluation of AI agents more seamless.

AIP Evals sidebar integration in Agent Studio

Users can now access AIP Evals directly from Agent Studio through a new sidebar integration. This allows creating evaluation suites, editing test cases and evaluators, running evaluations, and viewing results without leaving the application.

Integrate AIP Evals sidebar in Agent studio

The Metric Card widget now supports displaying object properties directly by inputting an object set with a single object and selecting a property, simplifying Metric Card configuration for builders. 

Display object properties directly with the Metric Card widget

Data Lineage

[Virtual tables](/foundry/data-integration/virtual-tables) will no longer appear with a grey header. Instead, use the new Resource type and Storage node coloring options to understand table types and external storage locations respectively. See [node color documentation](/foundry/data-lineage/node-coloring) for more information.

Virtual tables node styling changes

Foundry Branching

You can now protect the main branch of your Workshop modules and define custom approval policies. While this only applies to Workshop for now, all types of resources will eventually be supported, with support for ontology and Pipeline Builder resources coming next.

## Protect the main branch of your resources and define approval policies for your projects

You can now protect the main branch of your Workshop modules and define custom approval policies. While this only applies to Workshop for now, all types of resources will eventually be supported, with support for ontology and Pipeline Builder resources coming next. 

To safeguard critical workflows and maintain development best practices, you can protect the main branch of your resources. This means that any change to a protected resource must be made on a branch and will require approval to take effect.

![Approval Flow in a protected Workshop application.](./media/2025-07-09-111903-foundry-branching-approve-workshop-change-pn.png)

_Approval Flow in a protected Workshop application._

Once a resource is protected, any change to that resource will have to be made on a branch and go through an approval process. The approval policy is set at the project level, and defines whose approval is required in order to merge changes to protected resources.

![Project with default approval policy.](./media/2025-07-09-111907-foundry-branching-project-with-default-approval-policy-pn.png)

_Project with default approval policy._

![Project with custom approval policy.](./media/2025-07-09-111910-foundry-branching-project-with-custom-approval-policy-pn.png)

_Project with custom approval policy._

Approval policies have three customizable parameters:

 * **Eligible reviewers:** Define the users or groups that are allowed to review and approve changes to the main branch of a protected resource.
 * **Number of approvals required:** Define the minimum number of approvals needed to enable merging of a change. Options include any eligible reviewer, all eligible reviewers, or custom (specify the number of eligible reviewers).
 * **Additional requirements:** Control whether reviewers can approve changes to files they have contributed to in the proposed branch. A contributor is defined as any user who has made a change to that resource on the branch.

Note that branch protection currently only applies to Workshop resources, but support for protecting ontology resources is coming soon.




Protect the main branch of your resources and define approval policies for your projects

Ontology SDK

You can now create React applications directly from the **Developer Tools** page in Code Repositories using the new React template option. This addition allows you to quickly start building React applications with Ontology SDK integration, streamlining the development process for custom applications that interact with your Ontology data or the Foundry API.

React templates now available in Developer Tools

Workflow Builder

Workflow Builder now displays warning notifications when you navigate to it via a URL that doesn't add any content to your graph. This provides clear feedback in situations where previously there was no visual indication of why nothing appeared. These notifications appear when opening interfaces with no implementations, empty type groups, or function repositories without functions.

Warning notifications for empty graph navigation in Workflow Builder

AIP Evals now supports combining multiple object sets and manual test cases within a single evaluation suite. The test case creation experience has been simplified, allowing you to add, delete, and duplicate object sets as needed. This flexibility enables you to leverage object sets while also adding specific manual test cases for comprehensive function testing.

## Combine multiple object sets and manual test cases in AIP evaluation suites

AIP Evals now supports combining multiple object sets and manual test cases within a single evaluation suite. The test case creation experience has been simplified, allowing you to add, delete, and duplicate object sets as needed. This flexibility enables you to leverage object sets while also adding specific manual test cases for comprehensive function testing.

![You can now combine multiple object sets and manual test cases to an evaluation suite.](./media/2025-07-07-134908-aip-evals-aip-evals-create-suite-object-set-backed-test-cases-pn.png)

_You can now combine multiple object sets and manual test cases to an evaluation suite._

[Learn more about adding test cases in AIP Evals.](../aip-evals/create-suite.md#add-test-cases)

### We want to hear from you

As we continue to build upon AIP Evals, we want to hear about your experiences and welcome your feedback. Share your thoughts with Palantir Support channels or our [Developer Community ↗](https://community.palantir.com/).

Combine multiple object sets and manual test cases in AIP evaluation suites

You can now configure Boolean expressions directly within conditional blocks in AIP Logic. This enhancement eliminates the need to create separate transform blocks solely for use as conditions, streamlining your workflow when building conditional logic.

GA inline expressions in conditional transforms

The Markdown widget in Workshop now supports text tagging with the new Annotation feature. With this feature, builders can seamlessly display, create, and interact with annotation objects on text directly in the Markdown widget. 

## Annotate and tag text in Workshop with the Markdown widget

The Markdown widget in Workshop now supports text tagging with the new **Annotation** feature. With this feature, builders can seamlessly display, create, and interact with annotation objects on text directly in the Markdown widget. 

![An example of a Markdown widget with a configured create annotation action.](./media/2025-07-01-190308-workshop-widget-markdown-annotation-creation-pn.png)

_An example of a Markdown widget with a configured "create annotation" action._

Key highlights of this feature include: 

 * **Visual tags:** Display annotation objects as highlighted or underlined text with configurable colors.
 * **On-click interactions:** Users can interact with existing annotation objects in the widget by configuring actions and events. 
 * **User tagging:** Enable the creation of new annotation objects on specific portions of text.

![An example of a Markdown widget with configured annotation interactions.](./media/2025-07-01-190312-workshop-widget-markdown-action-on-annotation-pn.png)

_An example of a Markdown widget with configured annotation interactions._

To learn more about configuring **Annotations**, refer to the [Markdown widget documentation](../workshop/widgets-markdown.md).

### Your feedback matters

We want to hear about your experience with Workshop and welcome your feedback. Share your thoughts with Palantir Support channels or on our [Developer Community ↗](https://community.palantir.com/) using the [`workshop` tag ↗](https://community.palantir.com/tag/workshop).

Annotate and tag text in Workshop with the Markdown widget

AIP Agent Studio

Agent Studio view mode now allows users to toggle the visibility of sessions from all agent deployments in the session history panel. When opening a session, users can see exactly where it was created, including external applications, AIP Threads, Agent Studio, Functions, AIP Assist, or Workshop applications. This provides a central view for debugging agents without navigating between different applications.

View all agent sessions across deployments in Agent Studio

Users can now submit comments in the Comments widget using the CTRL/CMD + Enter keyboard shortcut. 

Submit comments in the Comments widget using CTRL/CMD + Enter 

A new webview is now available in the Palantir extension for Visual Studio Code's built-in source control panel. This webview displays code repository-specific actions such as the branch selector and the **Propose changes** option.



New repository webview is now available in the Palantir extension for Visual Studio Code

Data Lineage now supports copy-and-paste functionality. You can copy and paste notes as plain text, or use copy and paste to import a list of RIDs to Data Lineage or move nodes between tabs.

Copy and paste enabled for Data Lineage

Dataset preview in Data Lineage now displays data health information, including rules and health checks. This feature is available for all datasets, including streaming datasets.

Data health information now visible in dataset preview

When indexing an object on a branch, you can now choose whether to import user edits from the Main branch for that object type. This enables you to view objects and properties, including those created or modified by user input, directly on your branch.

Ability to index user edits on a branch

Modeling

Model Integration

Running experiments can now be tracked in the Models application. When experiments are created in a code workspace or code repository, they will automatically appear in the Models applications. This provides live metrics and charts for easier monitoring of long-running experiments.

Add support for viewing running experiments in Model app

Ontology Management

Ontology Manager now displays object view dependents for object types. You can see which managed Workshop modules are used in the latest published object view version, making it easier to understand dependencies between your object types and their views.

View object view dependents for object types in Ontology Manager

When looking at an interface in Ontology Manager, you can now open the interface in Workflow Builder using the **Open in** menu or by using the Ctrl+i (Windows) / Cmd+i (MacOS) keyboard shortcut, as with other resource types.

Easier access to Workflow Builder from Ontology Manager

The submission criteria copy-paste helper in Workflow Builder now includes a button that appears when hovering over any row in the table of actions. This button allows you to directly set an action as the source for copying submission criteria, providing an alternative to using the dropdown menu.

Direct selection button in submission criteria copy-paste helper

Workflow Builder now includes an option to filter search results to show only items from the currently selected Ontology. When enabled, searches for object types, action types, interfaces, and type groups will display results limited to the Ontology associated with your current workflow. This setting persists across workflows and sessions, reducing the need to scroll through results from multiple Ontologies.

Filter search results to current Ontology in Workflow Builder

When running an incremental transform, there are often situations where the entire input of a `SNAPSHOT` output needs to be read from the beginning, or one or more inputs to an incremental transform receives numerous transactions that collectively contain a lot of unprocessed data.

Typically, when an output dataset is built incrementally, all unprocessed transactions of each input dataset are processed in the same job. This job can take days to finish, often with no incremental progress. If the job fails halfway through, all progress is lost, and the output would need to be rebuilt. This process often results in undesirable costs and errors and does not address pipelines where large amounts of data need to be frequently processed. 

Limiting the maximum number of transactions that should be processed per job offers a solution to this time-consuming problem. Learn more in our [public documentation](/foundry/incremental-transaction-limits/).

## Limit batch size of incremental inputs to save time and compute costs

When running an incremental transform, you may encounter the following situations:

 * An output is built as a `SNAPSHOT` because the entire input needs to be read from the beginning (for example, the [semantic version](../transforms-python/incremental-reference.md#incremental-decorator) of the incremental transform was increased).
 * An output is built incrementally, but one or more inputs to the transform receive numerous [transactions](../data-integration/datasets.md#transactions) that collectively contain a lot of unprocessed data.

Typically, when an output dataset is built incrementally, all unprocessed transactions of each input dataset are processed in the same [job](../data-integration/builds.md#jobs-and-jobspecs). This job can take days to finish, often with no incremental progress. If the job fails halfway through, all progress is lost, and the output would need to be rebuilt. This process often results in undesirable costs and errors and does not address pipelines where large amounts of data need to be frequently processed.

Limiting the maximum number of transactions that should be processed per job offers a solution to this time-consuming problem.

![An animation of incremental transform builds. On the left the transform without transaction limits is constantly working on one job without noticeable progress. On the right the transform has set a transaction limit of 3 for the input and is progressing through jobs consistently.](./media/2025-07-01-135752-code-repositories-batching-incremental-gi.gif)

_An animation of incremental transform builds. On the left, the transform without transaction limits is constantly working on one job without noticeable progress. On the right, the transform has set a transaction limit of 3 for the input and is progressing through jobs consistently._

### Add transaction limits to inputs

If a transform and its inputs satisfy all [requirements](#requirements-and-limitations), you can configure each incremental input using the `transaction_limit` setting. Each input can be configured with a different limit. The example below configures an incremental transform to use the following:

 * Two incremental inputs, each with a different transaction limit
 * An incremental input that does not use a transaction limit
 * A snapshot input

```python
from transforms.api import transform, Input, Output, incremental

@incremental(
    v2_semantics=True,
    strict_append=True,
    snapshot_inputs=["snapshot_input"]
)
@transform(
    # Incremental input configured to read a maximum of 3 transactions
    input_1=Input("/examples/input_1", transaction_limit=3),

    # Incremental input configured to read a maximum of 2 transactions
    input_2=Input("/examples/input_2", transaction_limit=2),

    # Incremental input without a transaction limit
    input_3=Input("/examples/input_3"),

    # Snapshot input whose entire view is read each time
    snapshot_input=Input("/examples/input_4"),

    output=Output("/examples/output")
)
def compute(input_1, input_2, input_3, snapshot_input, output):
    ...
```

### Next steps and additional resources

After configuring your incremental transform with transaction limits, you can continue to configure and monitor your builds with the following features and tools:

 * [**Create a build schedule:**](../transforms-python/incremental-transaction-limits.md#create-a-build-schedule-to-keep-outputs-up-to-date) Configure a schedule in Data Lineage to build at a regular interval, and enable an option that ensures your data is never stale.

![Ensure your data is always up-to-date by configuring a build schedule.](./media/2025-07-01-135757-code-repositories-self-triggering-schedule-pn.png)

_Ensure your data is always up-to-date by configuring a build schedule._

 *  [**Verify job ranges:**](../transforms-python/incremental-transaction-limits.md#view-job-transaction-ranges) Review Spark details for your build jobs to verify the transaction limits read per input.

 * [**Learn about read ranges when transaction limits are set:**](../transforms-python/incremental-transaction-limits.md#understand-read-ranges-for-inputs-with-transaction-limits) Review how the  `added`, `current`, and `previous`  read ranges are used differently when incremental transforms are configured with and [without](../transforms-python/incremental-reference.md#incrementaltransforminput) transaction limits. 

### Requirements and limitations

To use transaction limits in an incremental transform, ensure you have access to the necessary tools and services and that the transforms and datasets meet the requirements below.

The transform must meet the following conditions:

 * The [incremental](../transforms-python/incremental-reference.md#incremental-decorator) decorator is used, and the `v2_semantics` argument is set to `True`.
 * It is configured to use Python transforms version `3.25.0` or higher. Configure a job with [module pinning](../code-repositories/module-pinning.md) to use a specific version of Python transforms.
 * It cannot be a [lightweight transform](../transforms-python/lightweight-examples.md#incremental-workflows).

Input datasets must meet the following conditions to be configured with a transaction limit:

 * It must be a [transactional dataset](../data-integration/datasets.md#transactions) input.
 * In the [current view of the dataset](../data-integration/datasets.md#dataset-views), the input must have only [`APPEND`](../data-integration/datasets.md#append)  transactions; however, the starting transaction can be a [`SNAPSHOT`](../data-integration/datasets.md#snapshot).
 * It cannot be a [snapshot input](../transforms-python/incremental-reference.md#snapshot-inputs).

### Your feedback matters

We want to hear about your experiences when configuring incremental transforms with transaction limits, and we welcome your feedback. Share your thoughts with Palantir Support channels or our [Developer Community ↗](./incremental-reference.md#incrementaltransforminput).





Limit batch size of incremental inputs to save time and compute costs

When building your pipeline, you may need to roll back a [dataset](/foundry/data-integration/datasets/) and all of its downstream dependents to an earlier version. The pipeline rollback feature allows you to revert back to a [transaction](/foundry/data-integration/datasets/#transactions/) of an upstream dataset. When performing a rollback, the data provenance of the upstream dataset transaction is used to identify its downstream datasets and their corresponding transactions to create a final pipeline rollback state. Typically, this process would require several steps to properly roll back each affected dataset. With pipeline rollback, this is reduced to a few simple steps discussed below, along with the ability to preview the final pipeline state before confirming and proceeding with the rollback. Pipeline rollback also ensures that the incrementality of your pipeline is preserved.As you set up your rollback, you can choose to exclude any downstream datasets; these datasets will remain unchanged as the pipeline is rolled back to the selected transaction. Learn more in our [public documentation](/foundry/data-lineage/pipeline-rollback/). 

## Roll back pipelines to a previous state in Data Lineage [Beta]

When building your pipeline, you may need to roll back a [dataset](../data-integration/datasets.md) and all of its downstream dependents to an earlier version. There can be many reason for this, including the following:

 * You identified a mistake in the logic required to build a dataset and need to revert it.
 * Incorrect data was pushed into your pipeline from an upstream source.
 * An outage occurred, and you want to quickly navigate back to an earlier state of your pipeline.

The pipeline rollback feature allows you to revert back to a [transaction](../data-integration/datasets.md#transactions) of an upstream dataset. When performing a rollback, the data provenance of the upstream dataset transaction is used to identify its downstream datasets and their corresponding transactions to create a final pipeline rollback state. Typically, this process would require several steps to properly roll back each affected dataset. With pipeline rollback, this is reduced to a few simple steps discussed below, along with the ability to preview the final pipeline state before confirming and proceeding with the rollback. Pipeline rollback also ensures that the incrementality of your pipeline is preserved.

As you set up your rollback, you can choose to exclude any downstream datasets; these datasets will remain unchanged as the pipeline is rolled back to the selected transaction.

This feature is currently in the [beta stage of development](../platform-overview/development-life-cycle.md), and functionality may change before it is generally available.

### Execute a pipeline rollback

1. Navigate to a [Data Lineage](../data-lineage/overview.md) graph containing the upstream dataset you would like to roll back.
2. Select the dataset in the graph. Then, from the branch selector at the top of the graph, select the branch on which you would like to perform the rollback.
3. Select **View node properties** in the panel on the right.

![The right editor panel in Data Lineage with the option to View node properties.](./media/2025-06-30-174508-data-lineage-click-rollback-pn.png)

_The right editor panel in Data Lineage, with the option to_ ***View node properties***_._

 4. Select **Actions**, then **Rollback**.

5. Under **Selected transaction**, choose the transaction to which you would like to roll back.

![An example of a selected transaction.](./media/2025-06-30-174512-data-lineage-select-transaction-pn.png)

_An example of a selected transaction._

After choosing the transaction, downstream datasets will automatically be found and the states they will revert to if the rollback is actioned will be displayed.

Resource types that are unable to be rolled back, including streaming datasets, media sets, and restricted views, will be displayed under the **unsupported resources** section. Transactional datasets on which you do not have `Edit` access will also be included in this list.

6. Select the timestamp under each dataset to navigate to the [**History** page](../dataset-preview/overview.md#history-tab) of the input, where the corresponding transaction will be highlighted.

![A list of datasets with timestamps of the builds.](./media/2025-06-30-174515-data-lineage-preview-rollback-panel-pn.png)

_A list of datasets with timestamps of the builds._

7. Select any datasets to exclude from the rollback by selecting **—** to the right of the dataset name. Once excluded from rollback, the dataset will appear in the **Datasets excluded from rollback** section.

![A list of datasets selected for rollback that you can exclude.](./media/2025-06-30-174518-data-lineage-exclude-from-rollback-pn.png)

_A list of datasets selected for rollback that you can exclude._

8. To add an excluded dataset back to the rollback, select **+** to the right of the dataset name.

![A dataset excluded from rollback that you can choose to add back.](./media/2025-06-30-174522-data-lineage-add-to-rollback-pn.png)

_A dataset excluded from rollback that you can choose to add back._

9. After finalizing the state of your desired rollback, select **Rollback**. A confirmation dialog will appear.

![A confirmation dialog confirming the rollback of five dataset transactions and incremental state resets of two datasets.](./media/2025-06-30-174525-data-lineage-rollback-confirmation-required-pn.png)

_A confirmation dialog confirming the rollback of five dataset transactions and incremental state resets of two datasets._

10. Enter the branch name as confirmation, then select **Confirm rollback** to proceed.

![A confirmation of seven successful dataset rollbacks.](./media/2025-06-30-174528-data-lineage-rollback-complete-pn.png)

_A confirmation of seven successful dataset rollbacks._

11. Once the rollback is complete, navigate to the **History** tab of the datasets and notice that the rolled back transaction is now crossed out, as shown below:

![An example of a dataset that was rolled back with the rolled back transaction crossed out.](./media/2025-06-30-174531-data-lineage-branch-pointer-updated-pn.png)

_An example of a dataset that was rolled back, with the rolled back transaction crossed out._

### Additional resources and support

To learn more about pipeline rollbacks, review our [public documentation](../data-lineage/pipeline-rollback.md). We also invite you to share your feedback and any questions you have with Palantir Support or our [Developer Community ↗](https://community.palantir.com/tag/data-lineage).





Roll back pipelines to a previous state in Data Lineage [Beta]

Debug outputs in AIP Logic can now be added without requiring an evaluation suite. These outputs can be captured in execution datasets, allowing you to use the debug data throughout the Palantir platform.

Logic - Enable execution dataset

The Button Group widget now supports usage of a string variable for the button description. Previously, only a static text value could be configured for button descriptions.

Button Group widget's description field may now be backed by a string variable

Dataset Preview

When importing files to a dataset with the **Import and validate** option, a schema editor dialog is shown when validation fails. In the dialog, you are able to select a button to undo the import with a transaction rollback. We renamed this button from **Cancel** to **Undo import** to clarify that the import already occurred and downstream consumers may have already processed the new files.

Improved labeling for schema editor dialog button

Users can now rename job groupings in Pipeline Builder to better organize and identify their pipeline jobs.

Rename job groupings in Pipeline Builder

Code repositories in Foundry Branching [beta] will now display a detailed deployability status in the branching bar. This deployability status indicates whether a branch can be merged to a given repository.

Deployability checks in Foundry Branching

Several applications (including Workshop, Actions, and Automate) now support version range dependencies for functions. Review the documentation on [Functions versioning](/foundry/functions/functions-versioning/) and [Version range dependencies](/foundry/functions/version-range-dependencies-for-functions/) for more details.



## Version range dependencies for functions

Previously, Foundry applications could only depend on functions at pinned versions. With the introduction of version range dependencies for functions, applications like Workshop, Actions, and Automate can now depend on functions at version ranges. This enables automatic upgrades at runtime, saving you time in your development cycle and providing a downtime-free upgrade experience for [deployed Python functions](../functions/python-functions-deployed.md).

![Configure a version range dependency for a function-backed widget in Workshop using the new Auto upgrade toggle.](./media/2025-06-25-213640-code-repositories-image-2-pn.png)

_Configure a version range dependency for a function-backed widget in Workshop using the new_ ***Auto upgrade*** _toggle._ 

[Review version range dependencies documentation to better understand feature behavior and considerations.](../functions/version-range-dependencies-for-functions.md)

### Versioning your functions

The introduction of version range dependencies makes proper release versioning more critical than ever. To help you in this process, new backward compatibility checks are now available when tagging your functions.

![Backward compatibility checks run automatically when you tag a new version of your functions in Code Repositories.](./media/2025-06-25-213646-code-repositories-image-4-pn.png)

_Backward compatibility checks run automatically when you tag a new version of your functions in Code Repositories._ 

Learn [how to properly version your functions](../functions/functions-versioning.md) to ensure that you provide your consumers with a stable and reliable experience.

### Your feedback matters

We want to hear about your experience with functions and welcome your feedback. Share your thoughts with Palantir Support channels or on our [Developer Community ↗](https://community.palantir.com/) using the [`functions` tag ↗](https://community.palantir.com/tag/functions).




Version range dependencies for functions

Python 3.12 support in Code Repositories and VS Code Workspaces is now available across all enrollments, adding Python startup time improvements, optimized comprehensions, reduced memory footprint, and other new features. This version is available across all Python environments, including during preview, checks, and builds for Python transform operations. To benefit from the latest performance improvements and essential security fixes, upgrade your repositories and set your Python environment to recommended version or explicitly select Python 3.12.



## Python 3.12 support in Code Repositories and VS Code Workspaces now available

Python 3.12 support in Code Repositories and VS Code Workspaces is now available across all enrollments, adding Python startup time improvements, optimized comprehensions, reduced memory footprint, and other new features. This version is available across all Python environments, including during preview, checks, and builds for Python transform operations.

Highlights from Python 3.12 include:

* **Faster startup:** Reduced interpreter initialization time (10-15% improvement in all Python process startup times)
* **Memory optimization:** Lower memory usage overall
* **Improved multithreading:** Enhanced performance with subinterpreters
* **Inlined comprehensions:** Significantly faster Python comprehensions

To benefit from the latest performance improvements and essential security fixes, upgrade your repositories and set your Python environment to **recommended version** or explicitly select Python 3.12.

Code Repositories and VS Code Workspaces now support Python 3.12

Python transforms that produce or consume models can now be previewed using the [Palantir extension for Visual Studio Code](/foundry/palantir-extension-for-visual-studio-code/overview/). Preview can be ran both in [VS Code Workspaces](/foundry/vs-code/overview/) in Foundry, or in local environments by [installing the extension](/foundry/palantir-extension-for-visual-studio-code/overview#1-download-the-palantir-extension).

Model preview now supported in  the Palantir extension for Visual Studio Code

An improved experience is now available when creating new or empty graphs in Workflow Builder. In addition to using the keyboard shortcut of `CMD + i` (MacOS) or `CTRL + i` (Windows) from a Workshop module, function repository, or, newly, an ontology object to jump into Workflow Builder, there is now another convenient way to select the most relevant Workshop modules directly from the application. From Workflow Builder's landing page, you can now see a list of the most recent or the most viewed Workshop modules. [Learn more about Workflow Builder.](/foundry/workflow-builder/overview/)



## Access the most relevant Workshop modules directly from the new Workflow Builder landing page 

An improved experience is now available when creating new or empty graphs in Workflow Builder. In addition to using the keyboard shortcut of `CMD + i` (MacOS) or `CTRL + i` (Windows) from a Workshop module, function repository, or, newly, an ontology object to jump into Workflow Builder, there is now another convenient way to select the most relevant Workshop modules directly from the application. From Workflow Builder's landing page, you can now see a list of the most recent or the most viewed Workshop modules.

![You can now access the most relevant Workshop modules directly from the new Workflow Builder landing page.](./media/2025-06-26-101621-workflow-builder-2025-06-25-213740-workflow-builder-screenshot-2025-06-18-at-3-47-53-pm-pn-pn.png)

_You can now access the most relevant Workshop modules directly from the new Workflow Builder landing page._ 

### What’s New?

When you open a new or empty graph, you will now see:

 * **The most recent Workshop applications:** Quickly access the most recently opened Workshop applications in your workspace.
 * **The most viewed Workshop applications:** Instantly discover the most popular and impactful workflows being used by your team. This option shows the most viewed Workshop applications in the last 30 days by all users.

![Select from the most viewed Workshop modules.](./media/2025-06-26-101624-workflow-builder-2025-06-25-213745-workflow-builder-screenshot-2025-06-18-at-3-48-16-pm-pn-pn.png)

_Select from the most viewed Workshop modules._

When you select a Workshop application, Workflow Builder will automatically populate the Workflow Builder graph with the backing resources, similar to how it populates the graph when a user uses the `CMD + i` keyboard combination directly from a Workshop module.

This update makes it easier than ever to get started: Explore, learn, and build with the most important workflows right at your fingertips.

[Learn more about Workflow Builder features in the documentation.](../workflow-builder/overview.md)

### We want to hear from you

Share your feedback with us through Palantir Support or our [Developer Community ↗](https://community.palantir.com/) using the [`workflow-builder` tag ↗](https://community.palantir.com/tag/workflow-builder).


Access the most relevant Workshop modules directly from the new Workflow Builder landing page 

Control Panel

Organizations can now be renamed in Control Panel. To use this feature, navigate to the **Organization management**page to find the organization list. Choose the organization you wish to rename, then select **Actions > Rename organization**. Note that this action is only available for users with the `Enrollment Administrator` role.

Rename organizations in Control Panel

Walkthroughs

Walkthrough builders can now configure highlighting of Workshop modules from within a Carbon module. To configure this, add the Carbon module as a step's primary resource and add any additional Workshop modules that you want to highlight.


Highlight Workshop modules within a Carbon workspace

The AIP Interactive widget now supports a configuration toggle that determines whether the session history panel shows sessions from all agent usages. This includes sessions from external applications, AIP Threads, Agent Studio, Functions, AIP Assist, and additional Workshop applications. With this feature, users can more easily transition between mobile and desktop workshop modules when chatting with an agent.

Show all sessions across usages in AIP Interactive widget

You can now import and use query functions in your TypeScript functions in Code Repositories. You can add your query functions in the sidebar and call them directly from your code.

Query functions in TypeScript functions now generally available

Data Connection

Data Connection now allows you to duplicate existing webhooks. When you duplicate a webhook, all configuration fields except for the API name are copied to the new webhook. This functionality makes it easier to create multiple similar webhooks without having to manually recreate each configuration from scratch.

Duplicate webhook functionality

Ontology Manager now supports the `Ctrl+I` keyboard shortcut to open the currently selected object type, action type, or type group in Workflow Builder. This shortcut provides a faster alternative to using the existing buttons for the same functionality.

Keyboard shortcut to open items in Workflow Builder 

 Workshop modules now display a flashing animation when highlighted by a Walkthrough. This visual enhancement makes it easier for users to identify and follow the highlighted elements during tutorial steps.


Animation for Workshop module highlighting

Solution Design Application

Solution Designer now autosaves your work regularly. Autosaved diagrams are visible only to you until you manually save them using the "Save" button (top right). You can discard the current autosaved version at any time, similar to other applications. Your diagram must be saved as a resource in a folder to be autosaved.

Autosave diagrams in Solution Designer

The `Toggle between light and dark theme` event will now work for Workshop modules viewed through a Carbon workspace configured to display in dark theme. Previously if a Carbon workspace was presented in dark theme, this event would not temporarily override the theme to light theme. Additionally, the toggled theme will now apply to the full Carbon workspace, rather than just the individual Workshop module.

Fix Workshop toggle theme event in dark theme Carbon workspace

The Object View widget now includes a direct link to Ontology Manager for the currently displayed object type. This allows builders to quickly access and edit the object type's configuration and Object View without having to manually search for the object type.

Link to Ontology Manager added in Object View widget 

Data Lineage hotkeys are being updated for better consistency. The **Select parents** and **Select children** actions now use left/right arrow keys without requiring modifier keys. Additionally, presentation mode hotkeys have been simplified. Modified hotkeys now work only with the documented modifier key (`Cmd` on macOS, `Ctrl` on Windows/Linux) rather than accepting any modifier. During the transition period, both old and new combinations will work, with a one-time warning displayed when using deprecated combinations.

Hotkey updates for improved consistency

Scheduler

The **Started by** column on the **Schedule Metrics** page now displays the fulfilled triggers for each schedule run, showing related builds or logic updates.

Scheduler run reasons

Streaming

Stream monitors can now be configured to monitor a specific branch. When creating or editing a monitor, you can specify which branch to monitor, and the monitor will only create alerts for that particular branch. This provides more targeted monitoring capabilities for streams with multiple branches.

Branch-specific monitoring for streams

The cleanup page in Ontology Manager now includes an open in Workflow Builder button, matching the functionality already available on the object types page. This addition allows users to quickly navigate from the cleanup page directly to Workflow Builder.

Open in Workflow Builder button added to cleanup page in Ontology Manager

The rule summary graph for actions and functions in Ontology Manager has been redesigned to improve readability. The new layout provides a clearer visualization of rule relationships and dependencies.

Improved rule summary graph readability in Ontology Manager

When creating a new workflow in Workflow Builder and selecting a workshop from the landing page, the system now automatically expands to show all dependencies of that workshop. This provides the same result as manually using Ctrl+I from within the workshop itself, giving you immediate visibility into the complete workflow structure.

Automatically display Workshop application dependencies

Automate

To enable users to view telemetry for their automations, a new telemetry tab has been added to automation overviews. In the telemetry tab, you can aggregate and monitor various metrics over recent executions. Note that the telemetry tab is only available for project-scoped automations.

Telemetry tab for Automate

Organizations can now be renamed in Control Panel by those with the `Enrollment administrator` role for the enrollment associated with the organization. To rename an organization, navigate to the **Organization management** page in Control Panel. Then select **Actions -> Rename organization**. To avoid breaking changes going forward,  ensure all integrations in the Palantir platform (such as OSDK and Code Repositories) are updated to use the organization RID instead.

Organizations can now be renamed in Control Panel

Metric objectives are now available in AIP Evals, giving you more control over how your evaluation results are measured and interpreted. Previously, users could see raw metric values for each test case. However, it was difficult to determine at a glance whether a model’s output truly met expectations especially when dealing with multiple metrics or large suites. With metric objectives, you can now define what success looks like for each metric in your evaluation suite. For Boolean metrics, you can choose whether true or false indicates a passing result. For numeric metrics, you can select whether higher or lower values are better, and optionally set a threshold for passing. This makes it easier to enforce standards and quickly identify regressions or improvements in your models. [Learn more about metric objectives in AIP Evals](/foundry/aip-evals/create-suite#evaluators).

## Introducing metric objectives for AIP Evals

Metric objectives are now available in AIP Evals, giving you more control over how your evaluation results are measured and interpreted. Previously, users could see raw metric values for each test case. However, it was difficult to determine at a glance whether a model’s output truly met expectations, especially when dealing with multiple metrics or large suites.

With metric objectives, you can now define what success looks like for each metric in your evaluation suite. For Boolean metrics, you can choose whether `true` or `false` indicates a passing result. For numeric metrics, you can select whether higher or lower values are better, and optionally set a threshold for passing. This makes it easier to enforce standards and quickly identify regressions or improvements in your models.

![Pass / fail configurations configured for Boolean and numerical metrics.](./media/2025-06-19-101201-aip-evals-image-pn.png)

_Pass / fail configurations configured for Boolean and numerical metrics._

When you run an evaluation suite, AIP Evals will automatically determine and display the pass/fail status for each metric and test case based on your configured objectives. The run results dialog now clearly visualizes which metrics and test cases passed or failed according to your criteria, making it easier to track quality and communicate results.


![Pass/fail status for metrics and test cases in the run results dialog.](./media/2025-06-19-101205-aip-evals-image-pn.png)

_Pass/fail status for metrics and test cases in the run results dialog._

[Learn more about metric objectives in AIP Evals.](../aip-evals/create-suite.md#evaluators)

Introducing Metric Objectives for AIP Evals

The advanced to-do application, available through VS Code workspaces, demonstrates how to use advanced features of the Ontology SDK (OSDK) by building a to-do application as a practical example. You will learn about many OSDK features, including ontology interfaces, media content handling with media sets, runtime-derived properties, and platform SDKs. To access and install the advanced to-do application example, open the Code Workspaces application and choose to create a **+ New Workspace**. Then, select **VS Code  >  Applications** and find the application in the **curated examples list.** You can also search for the application in the platform **Examples (Build with AIP)**, or by selecting the **Ontology SDK reference examples** option in Developer Console.



## Build an advanced to-do application with OSDK, now available in VS Code workspaces

The **advanced to-do application**, available through **VS Code workspaces**, demonstrates how to use advanced features of the Ontology SDK (OSDK) and platform SDK by building a to-do application as a practical example. You will learn about many OSDK features, including the following:

 * [**Ontology interfaces**](../interfaces/interface-overview.md)**:** Use ontology interfaces to interact seamlessly with backend services.
 * [**Media content handling with media sets**](../data-integration/media-sets.md)**:** Handle media content efficiently within your to-do application.
 * [**Runtime-derived properties**](../ontology/derived-properties.md)**:** Leverage runtime-derived properties for dynamic data handling and real-time updates.
 * [**Platform SDKs**](../dev-toolchain/overview.md#platform-sdks)**:** Integrate with the platform SDK for enhanced application functionality.

![The advanced to-do application with annotations where advanced OSDK features are used.](./media/2025-06-24-120206-ontology-sdk-advanced-to-do-application-pn.png)

_The advanced to-do application, with annotations where advanced OSDK features are used._

### Get started

To access and install the advanced to-do application example, open the Code Workspaces application and choose to create a **+ New Workspace**. Then, select **VS Code > Applications** and find the application in the **curated examples list.** 

You can also search for the application in the platform **Examples (Build with AIP)**, or by selecting the **Ontology SDK reference examples** option in Developer Console.

![Create an application in a VS Code workspace to choose the advanced to-do application example.](./media/2025-06-24-120210-ontology-sdk-screenshot-2025-06-23-at-14-04-44-pn.png)

_Create an application in a VS Code workspace to choose the advanced to-do application example._

The advanced to-do application application will then be installed through [Marketplace](../marketplace/overview.md).

### What's included in this example?

When you first install the example, it will include the `Advanced to-do application ontology` , an ontology that can be installed only once in each space to ensure API names remain unique and constant. After installation, be sure that all objects were synced to the ontology before you run the application; you can verify this by opening the `Advanced to-do`  object types in Object Explorer. After syncing, the data from the ontology should automatically appear when you open the example.

The installed example will open the code repository and run the application in a [VS Code workspace](../vs-code/overview.md) using the Palantir extension for Visual Studio Code. Review the included Markdown documents to learn more about the project and detailed explanations for data services. You can also [explore our public documentation](../ontology-sdk-react-applications/advanced-todo-application-overview.md) to learn more about the architecture and configuration of the application and the custom React hooks used. 

### What's next?

We are currently working on more examples to demonstrate other features available with the OSDK, including write operations with ontology interfaces and other advanced filtering patterns. Share your feedback with Palantir Support or our [Developer Community using the `ontology-sdk` tag ↗](https://community.palantir.com/tag/ontology-sdk) if you have ideas for more examples and tutorials that you would like to see.





Build an advanced to-do application with OSDK, now available in VS Code workspaces

Resource Management

The project and ontology dropdown filters in the **Resource Management Analysis** tab now correctly display options based on the selected invoiced dimension. Previously, these filters would only show options based on the first invoiced dimension, which could result in missing relevant options or showing irrelevant ones. This improvement ensures that the dropdown options consistently match the data displayed in the table and chart.

Fixed project and ontology filtering in Resource Management Analysis tab

Fixed an issue where iframing another part of the Foundry workspace within a Workshop module could sometimes cause a walkthrough to appear twice.

Fixed duplicate walkthroughs in Workshop modules

AIP Logic now supports media chaining, allowing you to connect multiple media expressions together in sequence. This feature enables more complex media operations workflows while maintaining chain integrity even when expressions are removed. Media expressions that don't output media will automatically end the chain.


[Media chaining] remove feature flag

AIP Logic now displays a warning when timestamps are referenced directly in prompts, helping prevent misinterpretation by LLMs. A  button is provided to automatically add a formatting block, ensuring timestamps are properly formatted for reliable processing.

Add auto-fixer for timestamps in prompts

Workshop's XY Chart widget now correctly applies ontology formatting across all layers. Previously, formatting settings might not have been consistently applied to all layers in the chart

Fix ontology formatting across layers in Workshop's XY Chart widget

Stream monitoring now includes utilization warnings that alert you when compute resources exceed moderate (80%) or critical (120%) thresholds. These warnings automatically prioritize affected charts by moving them to the top of the page. A new summary of all utilization warnings appears directly on the stream page, eliminating the need to navigate to the metrics page. When warnings occur, you can quickly upgrade your compute profile using a direct link to the pipeline builder's build settings.

Utilization warnings for stream compute resources

Foundry Branching now renders markdown formatting in proposal descriptions. This allows for better text formatting and interactive hyperlinks, improving the readability of proposal descriptions.

Markdown support in Foundry Branching pull request descriptions

Users can now configure the allocation of virtual CPUs (vCPUs) for serverless functions. This capability allows for better resource management and performance optimization based on specific function requirements. To configure vCPU allocation for serverless functions, navigate to Ontology Manager, select the relevant function and function version, then save the changes under the **Configuration** tab.

Configure vCPU allocation for serverless functions

Upgrade Assistant has added a new heuristic called "needs triage" for assigning resources to users to be actioned. "Needs triage" is applied in cases where standard heuristics are not sufficient. For example, these may be cases where there would otherwise be no assignees identified, or when the number of users with the project owner role is very large. See [resource assignment documentation](/foundry/upgrade-assistant/resource-assignment/) for more details.

New "needs triage" heuristic for resource assignment in Upgrade Assistant 

The Upgrade Assistant application homepage now has a new, more intuitive design, with clearer separation of active, past due, and archived upgrades through tabs. Filters and upgrade details are now more prominently displayed, along with more context such as upgrade-specific descriptive text and days remaining until the due date.

Updated homepage for Upgrade Assistant

The Object View widget now supports module interface mapping, allowing variables to be passed from the parent module to the object view. Existing Object View widgets need to be upgraded in the widget config panel to access this feature. To learn more about Object View widget interface configuration, review the [product documentation](/foundry/workshop/widgets-object-view/).

Module interface mapping is now supported in the Object View widget

The Tabs widget now better identifies the selected tab if multiple layout events are configured, such as in nested tabs setups. The selected tab will now be identified based on the number of configured events which match the current layout state. Previously, the first tab with any event matching the current layout state would be identified as selected.

Improved Tabs widget selection with multiple layout events 

When configuring the Button Group widget to open a URL, a new **Open URL in the current tab** toggle may be enabled giving builders the option to set whether a URL is opened in the current tab or a new tab. 

New open in current tab toggle in Button Group widget

Until now, the Data Connection app treated an export with "nothing new to export" as a build failure. From now on, exports with "nothing new to export" will be considered to be successful builds. This change is meant to disambiguate these "nothing new" exports from cases where an export fails because of an issue that the user must address or remediate. Treating a "nothing new" export as a success more accurately represents the reality that the export is not broken and there are no issues that need to be addressed by the user. 

This change may impact workflows that depend on the original failure behavior; users will need to adjust affected workflows accordingly.

Exports with nothing new are no longer considered build failures

The Data Lineage application now offers node coloring by compute engine. This enables viewers of data lineage to understand which compute engine drove the compute, including both Palantir compute flavors (such as Spark, Flink, Ontology) and external compute engines (such as Databricks, Snowflake, BigQuery). 

Data Lineage now supports coloring nodes by compute engine

Pipeline Builder now offers an in-memory checkpoint strategy option. When you mark a node as a checkpoint, you can choose to persist the results in memory rather than using the default short-term disk storage. Access this option by right-clicking on a checkpoint node and selecting the checkpoint strategies dialog.

In-memory checkpoint strategy available in Pipeline Builder

Pipeline Builder now uses a faster validation system that improves performance when working with large pipelines and complex data types. This enhancement makes error checking more responsive and reduces loading times when validating pipelines and reviewing proposals.

Improved performance for large pipeline validation 

Ontology Manager now supports migrating object types from Object Storage V1 to Object Storage V2 while preserving user edit history. Enable the toggle to include edit history in your migration settings, and the history will be migrated to Object Storage V2. Edit history will also be enabled for the Object Storage V2 object type. More information is available in the [OSv1 to OSv2 migration documentation](/foundry/object-backend/osv1-osv2-migration/).

Release notes

loading...