Release Summary 25.16 | Aug 07, 2025
The following key features and improvements, along with bug fixes, have been released in Algonomy CXP products in the release version 25.16.
Recommend
Improved Deduplication Control for Email Recommendations
We've improved the way deduplication works across multiple placements in email recommendations. Previously, the order in which placements were filled could vary, sometimes resulting in inconsistent product exclusions.
With this update, deduplication now consistently follows the order defined by the dedupWith parameter. This ensures more predictable behavior and gives customers greater control over which products appear in each placement—leading to more reliable and effective email campaigns.
Jira: ENG-30699
Margin Sorting Now Available Across All Models in Configurable Strategies
The "Margin" sorting option is now supported across all models in Configurable Strategies, allowing teams to prioritize higher-margin products more effectively. Previously limited to a few models, this enhancement brings consistency and greater flexibility to margin-based optimization.
Margin sorting is enabled only if the relevant site configuration (margin cp model or profit cp model) is active. If not, the option appears disabled with a tooltip guiding users on how to activate it through model settings and product catalog updates.
Jira: ENG-30525
Ensemble AI
Ensemble AI: Shopper Experience Enhancements
Ensemble AI now offers a more refined shopping experience with updates to how similar products and swaps are displayed. Swap options can be set to show always, on hover, or be hidden, and swapped items now clearly show the original product for context.
The similar products UI has improved spacing and alignment, with mobile views displaying them as a layover for easier interaction. Additional controls include scroll support and options to set the number of lines for product labels, enhancing flexibility and usability across devices.
Jira: ENG-30616
Find
Improved Matching Accuracy in Hybrid search with Smarter Results Filtering
We’ve enhanced our product matching engine in Hybrid search to deliver more accurate and relevant results. Previously, the system returned all potential matches—even low-quality ones—leading to inconsistent outcomes. With our latest update, only high-confidence matches are shown, thanks to the introduction of a new relevance threshold. This ensures cleaner, more precise results without compromising performance—even for customers with large catalogs.
Jira: ENG-30737
Engage
Reports Now Available in Guided Selling
Guided Selling experiences now include a built-in Reports tab, allowing users to view performance metrics without navigating away. The reporting interface includes start and end date pickers, channel filters, tables, and visual graphs, making it easier to monitor and optimize Guided Selling experiences.
Jira: ENG-30460
Algo Chatbot
Enhanced Chatbot Accuracy with Updated Help Docs and Expanded Coverage
The vector database powering the chatbot has been enhanced with two key updates to improve accuracy and coverage. First, documentation from Integration and Personalization Cloud has been added, enabling the chatbot to reference these materials and deliver more relevant responses to user queries.
Additionally, the help documentation is now automatically synced with the vector database on a weekly basis. This process identifies newly added or updated pages, generates embeddings, and removes outdated content to keep the database clean and current. Unchanged pages are skipped to optimize storage and performance, with detailed logging for traceability. Together, these improvements ensure more accurate, comprehensive, and reliable chatbot support.
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 25.16.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Recommend: Improved Rule Filtering in Advanced Merchandising Auto Evaluation |
The automatic rule evaluation for Advanced Merchandising now excludes disabled and expired rules during model preparation. Previously, including these rules could lead to errors and unnecessary load. Only active rules are now processed, with an exception for expired rules that start within the next day. This improves the efficiency and reliability of the evaluation process. |
07-Aug-25 |
|
|
Recommend: Improved Error Handling for Model Audits and Rule Imports
|
We’ve made model audits and Advanced Merchandising rule imports more resilient to errors. Model recommendation jobs now handle deleted DSW strategies gracefully, preventing audit failures and avoiding manual intervention. For rule imports, errors in individual rules no longer block the entire file. Problematic rules are skipped with clear messages, and duplicate placements are handled by selecting the active one—ensuring smoother, more reliable uploads. |
07-Aug-25 |
|
|
Chatbot: Authentication Added to Public Chat API Endpoints |
Authentication is now enforced for publicly exposed Chat API endpoints to enhance security and prevent unauthorized access. Users must authenticate before receiving a response, ensuring that access to the service is restricted to authorized users only. This applies to both QA and production environments. |
07-Aug-25 |
|
|
Chatbot: Feedback Mechanism Added to Chatbot Responses |
Users can now provide quick feedback on individual chatbot responses using thumbs up or thumbs down. If a response is marked with a thumbs down, the chatbot will prompt the user to share what could have been better, allowing for optional comments. |
07-Aug-25 |
|
|
Chatbot: Mobile Web Support for Chatbot UI |
The chatbot UI now supports mobile web, enabling users on phones and tablets to interact seamlessly with the chat experience. The updated design ensures products are displayed within the same area as questions and answers, using the full width of the device for better visibility. Users can scroll through the conversation and product suggestions easily, ensuring a smooth and consistent experience across devices. |
07-Aug-25 |
|
|
Recommend: Error Messages Now Shown for Advanced Merchandising Rule Imports |
When importing Advanced Merchandising rules via file upload, any errors returned by the system are now clearly displayed in a modal. The message includes specific details from the response, helping users quickly identify and resolve issues in the uploaded file. |
07-Aug-25 |
|
|
ThoughtSpot Reporting Enabled by Default |
ThoughtSpot (TS) reporting is now enabled by default for new sites. Previously, this setting had to be manually activated in the site dashboard configurations, which was often overlooked. This change removes the need for manual setup and ensures TS reports are readily available without additional configuration. |
07-Aug-25 |
|
|
Enterprise Dashboard: Category Picker Added to Configurable Strategies |
The category input in Configurable Strategies has been replaced with a more user-friendly category picker. This change addresses limitations with the previous autocomplete, making it easier to find and select categories. Users can now browse and select multiple categories for fixed seeds, while result views allow selecting one category at a time. |
07-Aug-25 |
|
|
Find: Vector Indexing Now Scheduled During Low-Traffic Hours |
To optimize performance and reduce load, vector indexing jobs for query and catalog data are now scheduled during each client’s lowest traffic periods. This change helps minimize the impact of resource-intensive operations like HNSW graph indexing, ensuring smoother overall system performance. |
07-Aug-25 |
|
|
Chatbot: Chatbot Performance Optimized Through Model and Prompt Tuning |
Chatbot performance has been improved by switching to a lighter OpenAI model and fine-tuning prompts. The system now uses the gpt-4-0-mini model to reduce response time and resource usage while maintaining answer quality. |
07-Aug-25 |
|
|
Universal Recipe Catalog Introduced for Cross-Sell Recommendations |
A centralized Universal Recipe Catalog has been implemented to recommend recipes based on seed products, supporting cross-sell strategies across all clients. A manually triggered job uses science team logic to generate recipes from a static list of seed ingredients, which includes metadata such as locale and language. The resulting recipe data and images are stored and made accessible via front-end requests. |
07-Aug-25 |
|
|
Unnecessary Logging Removed from Vector Jobs |
To reduce log file size and improve Azkaban usability, excessive logging (e.g., memory usage logs) has been removed from the following jobs: catalog-vector, query-vector, query-writer, and related-searches. |
07-Aug-25 |
Bug and Support Fixes
The following issues have been fixed in the release version 25.16.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
|
Corrected Parent ID Format for Category Ingestion with Force Flag |
When a category was ingested with the force flag enabled, the parent ID in the engine output was incorrectly written as a string instead of an array, causing NCA to drop the association. This has now been fixed — the parent ID is consistently output as an array for both force and non-force ingestions. |
07-Aug-25 |
|
MVT Test Details Page Fails to Load
|
Users were unable to view MVT Test Details because a deleted strategy rule caused the treatment modal to fail. The fix ensures that missing rules are ignored during loading. |
07-Aug-25 |
|
|
Social Proof: Incorrect Currency Symbol in SP Report |
The Social Proof report for a client incorrectly showed a $ symbol in the sales column. This has been fixed by removing the symbol and adding a separate currency column, consistent with Dynamic Experience Reporting. |
07-Aug-25 |
|
|
Recommend: Incorrect Error Status for Unknown Attribute in Recs API |
The API previously returned a status of “error” when an unknown attribute was used as a refinement, even though recommendations were available. This broke the Recs Test Drive UI. The issue is now fixed. Valid recs no longer return an error status, and the Rules tab displays as expected. |
07-Aug-25 |
|
|
Incorrect Search Results for “Shirts” Query
|
We’ve enhanced our search experience to ensure better accuracy. Searches for “women shirts” will now return only the most relevant results, excluding unrelated items like “t-shirts.” This improvement helps you find exactly what you’re looking for, faster and more efficiently. |
07-Aug-25 |