General Enhancements and Changes

Summary

The following enhancements and changes have been made as part of the 2026.2 update:

  • The STEP Workbench has been updated to support Apple Silicon, ensuring macOS users maintain uninterrupted access to the workbench as Apple phases out support for Intel-based applications. Workbench Launchers are now available for macOS on Apple computers that use Apple Silicon chips or Intel processors. To determine which Workbench Launcher is required, users can check their computer's chip type by navigating in their macOS to Apple menu > About This Mac.

  • The Execution Report section for all background processes now consistently displays the ISO date / time stamp before the informational log entry. This change makes it easier to trace execution times during troubleshooting.

  • New JavaScript logic has been introduced for ETIM, ECLASS, BMECAT, and Automotive customers which allows business rules to perform recursive changes to the dataset. Business rules can now be used to automatically remove all products from a classification hierarchy based on specific attribute values. Additionally, attribute groups, including all of their children, as well as object types, can be deleted in the same manner.

  • A new plugin called 'eClass Business Action Mapping' can now be utilized when applying autogenerated content to products. The plugin provides the user with the ability to apply advanced logic and complex automated structure to products using business rules.

  • A link to the 'Product Updates Center' is now available from the Start Page, and allows users to discover the latest innovations, enhancements, and technical updates across Stibo Systems.

  • The Elasticsearch reindexing process no longer occupies shared workers, which allows reindexing to be faster in many cases. Additionally, the batch size of indexing processes spanned by reindexing is increased.

These enhancements and changes are also included in the 2026.2 update and are described in the Details section that follows:

  • Search in the New UI and faceted search in the Web UI, both powered by Elasticsearch, now support querying text within assets. This functionality is available in ramp-up status for early adopters.

  •  Asynchronous translation updates offer improved flexibility in content included in translation files, as well as better integration with AI.

  • STEP now allows using SMTPS (Simple Mail Transfer Protocol Secure) for mail server communication, aligning with current security standards to ensure encrypted, authenticated email transmission.

  • Improvements to the One Queue background process (BGP) execution mechanism include dynamic scaling for background processes supporting multithreading to boost throughput and reduce CPU overload, as well as optimizations for InDesign Publishing and Async Translation BGPs.

  • The new Consumer Model Context Protocol (MCP) Server enables AI agents to access approved master data through a standardized, governed, and discoverable interface. Building on the capabilities of Data as a Service (DaaS), it provides an AI-friendly semantic layer that helps agents understand, query, and use trusted enterprise data consistently across multiple AI-driven scenarios.

  • Filter persistence behavior in Web UI tables can now be controlled through configuration.

  • Changes to how admins integrate PDX with STEP have been made to simplify and streamline the process.

  • The background process (BGP) quarantine mechanism has been improved to prevent incorrect escalation after planned maintenance, and execution reports now provide clearer diagnostic information about why a process was stopped or restarted.

  • The Create dialog for the Globally Configured Multi Edit Data Container now supports a node picker for reference fields, improving usability when creating data containers with references.

Details

Full-text search within asset contents

Search in the New UI and faceted search in the Web UI now support full-text search on asset contents, powered by Elasticsearch, enabling users to search within the content of indexed assets — not just their attributes. When a query matches text inside an asset, that asset appears in search results alongside other objects indexed in the Elasticsearch configuration. This makes it faster and more efficient to locate relevant assets, particularly in environments that manage large volumes of documents.

The full-text search on asset content supports a wide range of file types, including PDF, Word, Excel, PowerPoint, plain text, HTML, CSV, and several other common document formats / MIME types. Archive formats such as .ZIP, audio, and video files are not supported. Additionally, asset contents must contain native or pre-processed text — OCR scanning is not performed on scanned documents or image-based files.

When asset content searching is enabled, a triggering definition for asset.uploaded is automatically added in the relevant event processor configuration to ensure that the contents of indexed assets are included correctly in searches.

Searchability limitations, including file size, character count, and dimension-dependent assets, are defined in the Searchable Object Types in Elasticsearch topic in the System Setup documentation documentation.

This functionality is in ramp-up status and can be enabled for early adopters. Contact your Stibo Systems representative for details. To learn more about the ramp-up phase / status, refer to the License and Component Lifecycle in the System Update and Patch Notes section of online help.

Asynchronous translation updates

With the 2026.2 update, three important improvements have been made to Stibo Systems' asynchronous translation capabilities that give data stewards and product managers additional tools to increase both the speed and accuracy of translations done using asynchronous translation. These updates are described below:

  • Glossaries: AI‑powered translations in STEP rely on language models to infer meaning and terminology from context. While effective for general language, this can result in inconsistent translation of customer‑specific terms such as brand names, technical vocabulary, and standard phrases. Without enforced terminology, reviewers must manually correct recurring issues to maintain brand and industry consistency. This increases review effort, slows translation workflows, and limits confidence when scaling AI translations across global catalogs. With 2026.2, glossary support is introduced for AI translations, enabling customers to define approved source‑to‑target term mappings that the AI must follow. Configured for asynchronous AI translation services in the STEP Workbench, these glossaries act as hard rules that enforce consistent terminology, resulting in AI translations that are more accurate and require fewer rounds of edits.

    For more information on this feature, refer to the Configuring an Asynchronous Translation Service topic.

  • Contextual reference data: Admin users now have a more granular ability to determine which reference metadata is sent for translations using asynchronous translation, whether for an AI translation service, or a traditional translation service. Previously, when a product was sent into asynchronous translation, all language- and non-language-dependent reference attribute metadata (for valid references) would be sent for translation, whether the data needed translation or not. Now, non-language-dependent reference metadata is excluded from translation by default.

    Further, admin users now have the ability to determine which reference metadata is sent into translation by either designating certain attributes as 'contextual,' or excluding the metadata attributes altogether. When designated as 'contextual,' the translation agency or AI translation service can use that reference metadata to better understand the meaning and intended usage of the content. This additional context helps guide translation choices, resulting in more accurate and consistent translations, delivered at no additional cost to the user. When excluded, the reference metadata attributes are not sent at all, reducing complexity, file sizes, and total translation costs.

    For more information on this feature, refer to the Attribute Filters for Asynchronous Translation Services topic.

  • Include or exclude object name: A new setting enables admin users to include or exclude, based on object type, a translated object's name when translating. This feature allows the user to:

    • Include the object name as translatable if the object name is defined as language-dependent in the object type

    • Include the object name in the translation as contextual

    • Exclude the object name from translations in all cases

    This setting applies globally to all object types and cannot be configured per object type.

    For more information on this feature, refer to the Attribute Filters for Asynchronous Translation Services topic.

These enhancements are delivered behind a configuration property and can be adopted at your own pace. Existing functionality will remain available for now but will be removed with a later update. The removal date has not yet been defined but will be communicated later. Contact your Account Manager or Customer Success Manager with any questions.

These changes provide users with a more robust and configurable asynchronous translation capability, empowering more dynamic integrations with AI, speeding translations through workflows, and getting usable translation data into the global market faster than ever before.

SMTPS support for mail server communication

STEP now supports SMTPS (Simple Mail Transfer Protocol Secure, also referred to as 'Implicit SSL/TLS') for mail server communication. This update aligns STEP with RFC 8314, the current industry standard that designates Cleartext email transmission as obsolete.

SMTPS enforces an encrypted connection from the start of each session, using a dedicated port (port 465). This approach ensures that email communications are authenticated, and that data integrity and confidentiality are maintained throughout transmission. System administrators responsible for mail server configuration will benefit most from this update, as it removes reliance on older, less secure protocols.

For more information, refer to the Email from STEP topic in the Resource Materials documentation.

One Queue introduces dynamic parallelism

By default, the One Queue background process (BGP) execution mechanism now dynamically scales parallel threads for 'DataProfiler Parallel’ and inbound integration endpoint (IIEP) background processes to manage fluctuating resource demand. (Other process types will be added in a future update.) This enhancement improves the processing speed for large workloads while maintaining fairness and stability across the platform. It maximizes the use of available processing capacity without manual tuning, reduces operational effort, and thereby supports predictable execution of business‑critical workloads. Additionally, BGPs for InDesign Publishing and Async Translation are optimized to work more efficiently with One Queue.

For more information, refer to the BGP One Queue topic in the System Setup documentation.

Consumer Model Context Protocol (MCP) Server

Agentic AI is increasingly used to answer questions, assist users, automate decisions, and support business processes. To deliver reliable outcomes, these agents need access to trusted, governed master data. However, connecting AI agents to master data has traditionally required custom integrations, APIs, or services that are often tightly coupled to specific use cases and underlying data implementations. While these integration approaches can support individual scenarios, they are typically designed for application-to-application communication rather than for AI agents that need to discover available data, understand its meaning, and access it safely. As AI adoption grows, this can lead to duplicated integration work, limited reuse, and increasing complexity around governance, scalability, and control.

With the 2026.2 update, Stibo Systems introduces the Consumer Model Context Protocol (MCP) Server. The Consumer MCP Server provides a standardized interaction layer between AI agents and approved master data. It enables agents to discover data, retrieve context, and query governed master data through a consistent and controlled interface. By building on Data as a Service (DaaS), the Consumer MCP Server adds an AI-friendly semantic layer that helps large language models and AI agents better understand, interpret, and act on enterprise master data. Instead of requiring hard-coded knowledge of the underlying data model, agents can interact with master data through a well-defined and discoverable interface. This makes trusted master data more accessible and reusable across AI-driven scenarios, including answering questions about products or customers, assisting users with contextual insights, and supporting internal or external AI assistants that rely on governed enterprise data.

For more information, refer to the 'Consumer MCP Server' section of the DaaS User Guide, available directly from the service.

Filter persistence behavior in Web UI tables can now be controlled through configuration

Filtering behavior in the Web UI has been improved to make persistent filters safer and easier to control when working with task lists and search results. You can now disable persistent filters across all Web UI tables using a shared configuration property.

This functionality is managed by a configuration property that is not available in the Self-Service UI. Contact Stibo Systems Support for assistance.

For more information on filtering in the Web UI, refer to the Node List Component topic in the Web User Interfaces documentation.

Improvements to PDX integration with STEP

In an effort to improve how admin users integrate PDX with STEP, a number of updates have been made as part of a larger project to make integration simpler. For the 2026.2 release, the following enhancements have been implemented:

  • Information in the background process's 'Execution Report' has been improved to help users quickly identify and correct errors related to authentication. Users are now provided with authentication error information in near real time.

  • Default values for integrations have been updated to better reflect recommended usage. Key defaults such as 'Number of Events to Batch' have been updated to align with platform expectations.  

  • Mandatory fields are now explicitly marked in the configuration to help users quickly identify which values must be provided.

These changes help streamline the integration process by improving configuration clarity, thus reducing the need for manual configuration changes.

For more information on integrating STEP and PDX specific to these changes, review the PDX Base Setup topic.

Improved BGP Quarantine Diagnostics

The background process (BGP) quarantine mechanism has been improved to prevent processes from being incorrectly escalated after planned maintenance. Restart counts now distinguish between crashes and intentional shutdowns, so that planned environment restarts no longer trigger quarantine.

Diagnostics are improved now that execution reports clearly state why a background process was stopped or restarted, including cases where a process was canceled because a long-running script exceeded its execution limit. Restart information is also logged more precisely, making it easier to identify which processes were affected after a crash or shutdown. These improvements make background process behavior easier to understand, troubleshoot, and operate.

For more information, refer to the BGP States and Quarantine Status topic in the System Setup documentation.

Node picker support in Globally Configured Multi Edit Data Container

Node pickers are now supported for references in the Create dialog of the Globally Configured Multi Edit Data Container. This enables faster and easier reference selection when creating data containers with references.

For more information, refer to the topic Globally Configured Multi Edit Data Container in the Web User Interfaces documentation.