Overall update summary
2025.4 is a self-service and scale release. Administrators gain role-based access control, sandbox backup and restore, and a range of new Self-Service UI capabilities that remove dependencies on the support team, alongside API access to security events and JavaScript logs. Data as a Service is strengthened on four fronts — caching, consumption control, attribute group mapping and performance. On top of that platform work, the release launches an AI-powered translation service and ProductGen AI asset analysis, makes Excel a more complete onboarding option, turns the Channel Portal into a self-service tool for retailers, redesigns the ECLASS ADVANCED data model for large volumes, and improves both Match & Merge performance and unmerge accuracy.
Log access — API access to security events and JavaScript logs
Platform | Included in Base | Admin, Developer | 25.4 highlight
What changed
2025.3 made security events available through Self-Service, but retrieving them was a manual process that did not support automated monitoring or auditing. With 2025.4, API access is introduced for both security events and JavaScript logs.
- Security events can be retrieved via API for automated monitoring and auditing.
- Streamlines security oversight.
- Removes a manual retrieval step.
Additional resources
Self-Service UI enhancements
Platform | Included in Base | Admin, Developer | 25.4 highlight | consolidates the deck's Self-Service UI 1/4–4/4 slides
What changed
The Self-Service UI supported only one admin role, so every user had full access to all functionality — a problem as features such as Log Access and the Modular Service Platform brought in different user personas. It also left several maintenance tasks dependent on the support team, and repeated use of Refresh Configuration for test preparation was cumbersome. With 2025.4, role-based access control becomes available when using External Authentication, sandbox database backups can be created and restored, and new capabilities are added: editing the STEP login screen welcome text and password reset options, creating an update preview sandbox from a preprod source, setting Scene7 and MongoDB configuration properties, validating whether active customizations or extensions are compatible with a new STEP version, filtering the delta changelog by component, and displaying platform alerts and service disruption notifications from StatusHub.
- Users access only the functionality relevant to their role, improving security and governance.
- Faster resets to a baseline state, useful for automated testing and quick environment refreshes.
- Greater self-service control, reducing reliance on support and improving operational efficiency.
Additional resources
Data as a Service enhancements
Platform | Included in the Data as a Service license | Admin, Developer | 25.4 highlight | consolidates the deck's Data as a Service 1/4–4/4 slides
What changed
Data as a Service processed repetitive requests individually without caching, gave customers no way to configure their own consumption limits, and relied primarily on metadata to control which attribute values a service exposed. 2025.4 addresses all three and adds a performance pass: request caching is introduced with control at both service and request level; request and object request limits become configurable at global, service or consumer level; service attributes can be mapped based on the STEP attribute groups they belong to; and updated libraries and components deliver performance and storage improvements.
- Significantly faster response times and reduced object consumption for repetitive requests.
- Helps prevent overconsumption, particularly in non-production environments during implementation and testing.
- Lets customers manage which attributes are exposed per service directly from STEP using attribute groups.
- Improved parallel processing and lower storage consumption, so more data fits within existing limits.
- Strengthens support for AI-driven use cases, complementing the Stibo Systems MCP Server offering (Limited Availability at the time of this release).
Additional resources
Product data maintenance in Excel
PXDC / Acquire & Share | Available with the PDX Onboarding license and the PDX Syndication license | Business User | 25.4 highlight
What changed
Many small suppliers prefer Excel for product data onboarding, especially for a handful of products, but Product Data Onboarding did not support attributes with a list of values in Excel files, making complex structures harder to manage. With 2025.4, users can import and export attributes' lists of values in Excel.
- Simplifies onboarding and reduces manual work.
- Makes Excel a more complete option for managing product data.
- Particularly helpful for small or non-technical users.
Additional resources
- Youtube chapter: Watch
Onboarding channel management for retailer admins
PXDC / Acquire | Available with the PDX Onboarding license | Admin | 25.4 highlight
What changed
With foundational structures for retailer and channel ownership and supplier registration in place in Product Data Onboarding, retailers needed an efficient, scalable self-service portal for managing onboarding channels. The 2025.4 update transforms the newly branded Channel Portal into a scalable self-service tool, including multichannel support and the migration of channel update jobs from Jenkins into Product Data Onboarding.
- Retailers manage onboarding channels themselves, more efficiently.
- Faster channel deployment.
- Prepares the platform for broader adoption.
Additional resources
- Youtube chapter: Watch
Improved seeded images management in channels
PXDC / Acquire | Available with the PDX Onboarding license | Admin | 25.4 highlight
What changed
When images retrieved from the Stibo Systems platform were resubmitted through Product Data Onboarding, new STEP IDs were generated — creating duplicate assets and breaking product links, because ID-generation rules intended to ensure uniqueness and support multiple asset object types had this unintended effect. With 2025.4, a mapping mechanism during maintenance and retrieval retains the original STEP ID for identical usages. On submission the existing asset is reused unless the usage type changes, in which case a new ID is generated.
- Prevents unnecessary asset duplication and maintains correct product-asset links.
- Still supports legitimate new usages.
- Repeated seeding and submission no longer create redundant assets, improving efficiency and data integrity.
Additional resources
- Youtube chapter: Watch
Scalable ECLASS ADVANCED data model
PXDC / Acquire, Manage, Share | Available with the ECLASS ADVANCED license | Admin, Business User, Developer | 25.4 highlight
What changed
The existing ECLASS ADVANCED data model did not scale effectively past roughly 1,000,000 products, creating performance bottlenecks and unmanageable workloads that risked delaying onboarding, enrichment and classification. The redesigned model in 2025.4 folds the entire ECLASS ADVANCED hierarchy into a single version-specific attribute on the product itself, represented internally in CSV format using IRDI as internal IDs.
- Reduces database and memory usage and simplifies the data structure.
- Improves import and export performance.
- Lets the platform scale efficiently at large product volumes.
Additional resources
Match & Merge — performance and scale
CXDC, BPDC, SDC / Acquire | Included in Base | Admin, Developer | 25.4 highlight
What changed
Using the Machine Learning Matcher in the Match & Merge Importer slowed import performance significantly, because invoking it was a blocking operation in the ranking process. With 2025.4, Machine Learning processes start concurrently.
- Using the Machine Learning Matcher no longer slows import performance.
- Higher throughput on Match & Merge imports.
- Removes a blocking step from the ranking process.
Additional resources
ProductGen AI — support for asset analysis
PXDC / Manage | Available with the ProductGen AI license | Admin, Business User, Developer | 25.4 highlight
What changed
Customers wanted to use generative AI to improve product data content, but doing so outside the MDM environment added complexity and risk and made safe experimentation at scale difficult. In 2025.4, the Azure OpenAI service lets customers bring images into generative AI workflows to generate descriptions from solution-specific prompts, discover exceptions to digital content requirements, and define data quality rules using natural language processing to determine what is and is not allowed.
- Streamlines content creation and quality control.
- Lets users analyze asset content alongside product data.
- Keeps generative AI experimentation inside the existing platform.
Additional resources
- Youtube chapter: Watch
AI-powered translations in the asynchronous framework
PXDC / Manage | Available with the Translation Service Integration and ProductGen AI licenses | Admin, Business User, Developer | 25.4 highlight
What changed
Enterprises handle large volumes of multilingual product content, but manual and agency-based translation is slow, expensive and hard to scale against omnichannel demands — delaying global updates and increasing costs. With 2025.4, a new AI-powered translation service launches within the asynchronous translation framework, integrating a Stibo Systems hosted large language model into the platform to automate multilingual translations.
- Enables multilingual updates across global markets.
- Reduces cost by minimizing reliance on translation agencies.
- Scales consistent, high-quality content creation across languages.
Additional resources
Unmerge improvements
CXDC, BPDC, SDC / Manage | Included in Base | Admin | 25.4 highlight
What changed
Survivorship rules do not rewrite data when incoming data is identical to what is already on the golden record, which caused problems for unmerge when a user was not unmerging the most recent merge — and the addition made in 2025.3 failed to restore multivalued references and data containers from previous revisions. With 2025.4, the unmerge logic is extended to read data from the deactivated golden record, or from the source record, to determine the right action.
- Correct data is restored where the same data versus no data was merged into a golden record.
- Multivalued references and data containers are restored from previous revisions.
- More reliable unmerge behavior beyond the most recent merge.
Additional resources
New UI improvements
PXDC / Manage | Please refer to your Account Manager | Business User | 25.4 highlight
What changed
As the New UI gains traction across PXDC customers, growing data model and business complexity means users expect the interface to adapt dynamically, validate inputs in real time, and be extensible to specific business needs. 2025.4 delivers three enhancements: dynamic conditional fields for attributes, reference types and data containers, giving context-driven validation and visibility; rendering of HTML functions within an embedded tab for flexible data presentation; and keyboard navigation for selecting and editing data without leaving the keyboard.
- Improved accuracy through context-aware validation.
- Richer data visualization through HTML rendering.
- Faster, more accessible data editing via keyboard navigation.
Additional resources
FAQ
2025.4 adds role-based access control, sandbox backup and restore, and a set of new capabilities to the Self-Service UI, plus API access to security events and JavaScript logs. Data as a Service gains request caching, configurable consumption limits, attribute group mapping and performance improvements. The release launches an AI-powered translation service and ProductGen AI asset analysis, adds Excel list-of-values support to onboarding, turns the Channel Portal into a self-service tool, fixes seeded image duplication, redesigns the ECLASS ADVANCED data model for scale, speeds up Match & Merge imports, improves unmerge accuracy, and adds dynamic conditional fields, HTML rendering and keyboard navigation to the New UI.
Administrators and developers are the main audience for the platform, Data as a Service, ECLASS ADVANCED and Match & Merge updates. Business users are targeted by Excel-based onboarding, ProductGen AI asset analysis, AI-powered translations and the New UI improvements.
Acquire covers onboarding channel management, seeded images and Match & Merge performance. Manage covers ProductGen AI asset analysis, AI-powered translations, unmerge improvements and the New UI. Product data maintenance in Excel spans Acquire and Share, and the ECLASS ADVANCED redesign spans Acquire, Manage and Share. Log access, the Self-Service UI and Data as a Service are platform-level.
Log access, Match & Merge performance and unmerge improvements are included in base, as are the Self-Service UI enhancements. Data as a Service enhancements are included in the Data as a Service license. Excel onboarding requires the PDX Onboarding and PDX Syndication licenses; channel management and seeded images require the PDX Onboarding license. ECLASS ADVANCED requires the ECLASS ADVANCED license, ProductGen AI asset analysis requires the ProductGen AI license, and AI-powered translations require the Translation Service Integration and ProductGen AI licenses. For the New UI improvements the deck directs customers to their Account Manager.
Links to the 2025.4 update notes, overview video, YouTube chapters and overview deck are listed under Supporting assets. [Insert / confirm]