26.04 Release notes

Highlights include workflow versioning, AI-powered document validation, PDF document translation, Python 3.13/3.14 support, R Package refresh, configurable inventory record types, enhanced permissions management, and the ​ValidMind MCP Server. This release also includes other enhancements and bug fixes across the platform.

Frontend

Workflow highlights

Workflow versioning and states

When you save changes to a workflow with active executions, a new version is now created automatically while the original is preserved. This enables you to:

• Browse version history and view read-only snapshots of previous versions.
• Add version notes when creating new versions.
• See which version each execution ran on in the execution details.

Active executions running on a previous version will continue to run on that version but can be restarted to run on the latest version.

Configure workflows

Versioning of workflows

User Action step renamed

The workflow step previously labeled User Action v2 is now simply called User Action. The legacy User Action step has been removed.

Workflow step types

User Action step in workflow editor

Record and artifact field and stakeholder variables in broadcast step

Broadcast step variable inputs are now based on the record or artifact type the workflow is tied to, including field values and stakeholders. For example, a Validation Issue workflow can reference Validation Issue fields and stakeholders, while a Use Case workflow can reference Use Case fields and stakeholders.

Workflow step types

Inserting variables into a broadcast step

AI-powered regulatory and policy alignment highlights

Document Checker for validation reports

Enable better regulatory and policy alignment by analyzing your validation reports against specific regulations or policies to identify gaps and improve compliance:

• Upload or select existing policy documents to check against.
• Run automated checks on validation report sections.
• Review detailed findings with specific recommendations.
• Generate a PDF report of Document Checker results.

Check documentation for compliance

Document Checker for validation reports

Testing automation highlights

Python 3.13 and 3.14 support

The ValidMind Library now supports Python 3.13 and 3.14, enabling you to use the latest Python features and performance improvements.

Refresh of ​ValidMind R Package

With this refresh, we're bringing the R package up to date with the latest ValidMind Library:

• Full API parity with the Python SDK for test execution and documentation.
• New quickstart notebooks demonstrating common R workflows.
• Updated documentation and examples for R users.

Install and initialize ​ValidMind for R

R Quickstart for Documentation

R Quickstart for Validation

​ValidMind R notebooks

Core documentation highlights

Numbered table and figure captions

Tables and figures within test results logged in documents now display numbered captions:

• Captions render as Table N or Figure N when no custom caption exists.
• Custom captions display as Table N. Caption or Figure N. Caption.
• Numbers automatically refresh when blocks are added, removed, or reordered.

Enable this feature in Settings > Organization > Document Defaults.

Work with test results

Numbered table and figure captions

AI-based automation highlights

Convert PDF uploads to ​ValidMind templates

Upload existing PDF documents and automatically translate them into ​ValidMind documentation format.

The document parser enables you to import existing model documentation into ValidMind Platform:

• Upload PDF files containing model documentation.
• Automatically extract text, tables, and structure from the PDF.
• Map extracted content to ​ValidMind document templates.
• Review and edit the imported content before finalizing.

Manage model documents

Import PDFs and convert to templates

Automatically map developer and validator evidence to validation reports

Analyze your model documentation and validation report to automatically map evidence:

• Identify relevant evidence from developer documentation.
• Suggest appropriate mappings to validation report sections.
• Reduce the manual effort required to link evidence across documents.
• Speed up the validation and review process.
• Keep the human in the loop for review.

Preparing validation reports

Automatically map evidence

Automatically generate evidence assessments in validation reports

Once evidence is linked to your validation report sections, you can auto-generate assessments that:

• Analyze the linked evidence content.
• Generate draft assessment text based on the evidence.
• Provide a starting point for validator review and refinement.
• Accelerate the documentation of validation findings.

Preparing validation reports

Auto-generate evidence assessments

Export Document Checker results as PDF

After running the Document Checker against your validation report, you can generate a PDF for offline review or compliance records that includes:

• Summary of compliance findings.
• Detailed results for each checked section.
• Recommendations for addressing gaps.
• A format suitable for audit documentation and stakeholder review.

Check documentation for compliance

Export Document Checker results

Inventory management highlights

Configurable inventory record types

Expand your governance tracking beyond models to any type of asset:

• Create custom record types for different governance needs like analytics models, attestations, or AI agents.
• Configure custom attributes specific to each record type.
• Assign dedicated documentation templates per record type.
• Set up independent workflow configurations for each type.
• Define relationships between different record types.

Managing the model inventory

Customizable inventory record types

Use model stage and stakeholders in calculated fields

Calculated fields now support additional inputs:

• Model stage: Use the current model stage value in calculations, enabling dynamic fields that update when models progress through lifecycle stages.
• Model stakeholders: Reference assigned stakeholder roles in formulas to calculate ownership metrics, review assignments, or role-based indicators.

Inventory field types

Calculated field inputs

Archive or delete artifact types

Remove artifact types that are no longer in use to keep your inventory clean:

• Archive artifact types to hide them from new artifact creation. Existing artifacts of default types will be also be hidden from view.
• Delete artifact types that have no associated artifacts.
• Review dependencies before deletion to understand impact.

Manage artifact types

Archive an artifact type

Platform capabilities highlights

​ValidMind MCP server

Connect AI assistants and developer tools to interact with the ValidMind Platform through text prompts via the Model Context Protocol (MCP):

• Connect Claude, Cursor, and other MCP-compatible tools to your ​ValidMind instance.
• Create and update models, model inventory fields, and documentation programmatically.
• Access model stages, use cases, stakeholders, and organizational data.
• Automate model governance tasks from your preferred development environment.

Connect AI assistants via MCP

MCP Server

Permissions settings for roles, stakeholders, and groups

New settings pages provide fine-grained control over permissions at the role, stakeholder, and group levels:

• Stakeholder permissions: Control access to model inventory fields, artifact fields, document types, and artifact types per stakeholder role. Search, filter, and bulk-edit permissions in a single view.
• Role settings: New layout for managing role-based permissions with improved visibility into what each role can access.
• Group settings: Manage members assigned to groups under a dedicated management page.

Manage permissions

Permissions settings

Model activity in exports of analytics data

Include model activity data when exporting analytics to S3 or S3-compatible storage, with schedule options and run history.

The analytics data export feature now supports model activity data:

• Export model activity alongside other analytics metrics.
• Choose a configurable 30, 60, or 90 day rolling window for activity data.
• Configure scheduled exports to S3 or S3-compatible storage.
• View export run history and status.
• Include activity data in your external reporting and data warehouse integrations.

Configure analytics exports

Model activity in exports of analytics data

API enhancements

Updated public API endpoints improve data export capabilities and align terminology with the rest of the platform:

• Findings renamed to artifacts: The API now uses artifacts terminology consistently, aligning with the user interface. Existing endpoints remain backwards-compatible during the transition period.
• Analytics widget exports: Export specific analytics dashboard widgets via the reporting API, enabling integration with external reporting tools.
• Model activity in data exports: Include model activity data when exporting analytics to S3 or compatible storage.

GET /artifacts GET /reporting/widgets/{widget_id}/metadata POST /reporting/widgets/{widget_id}/data

Artifacts in the Public API

Enhancements

Expired API key warnings

The Getting Started page now displays a warning when API keys are expired, helping users identify and refresh their credentials before encountering errors.

Install and initialize the ValidMind Library

Modify workflow execution end dates

Users with the Manage Workflow Execution permission can now modify the end date of running workflow executions, enabling adjustments to deadlines without stopping and restarting the workflow.

Configure workflows

Saved views scoped by record type

Saved views on the model inventory page are now scoped by primary record type. When switching between record types, only views relevant to the selected type are shown. New views are automatically associated with the currently selected record type.

Customize model inventory layout

Bug fixes & support

Fixed navigation issues in slow network environments

Fixed several navigation issues that could cause unexpected routing behavior, particularly in slow network environments. The loading spinner no longer unmounts the UI, async redirects can no longer override user navigation, and the sign-in callback no longer triggers unnecessary page reloads.

Fixed custom stakeholder type renaming

Restored the ability to rename custom model stakeholder types from Settings. The stakeholder detail page again supports inline editing of the display name.

Backend

Enhancements

SIEM bridge environment variable support

SIEM bridges now support configuration values from environment variables using the $-env syntax. This enables you to securely configure SIEM integration credentials without hardcoding them in the configuration file.

Configure AI model fallbacks in vmconfig.toml

An additional way to configure AI model settings is now available. In vmconfig.toml, you can specify model fallbacks for vm-flagship, vm-mini, and vm-embedding to customize which AI models are used for AI Content Builder and ValidChecker operations.

​ValidMind configuration reference (vmconfig.toml

4

Field-level access control for model inventory

Custom fields in the model inventory are now hidden from non-admin users unless their role has been explicitly granted read or write access. Previously, all users could see all fields in Manage Columns regardless of configured permissions.

Manage permissions

Bug fixes & support

Fixed aggregate field type filtering

Fixed an issue where the artifact aggregation field type filter was showing more field types than supported. Only date-type custom artifact fields are now available for aggregation filtering.

Fixed model stage renaming restriction

You can now rename model stages without restriction. Previously, renaming a model stage that was assigned to an inventory was more restrictive than necessary.