Release Summary 25.14 | Jul 10, 2025
The following key features and improvements, along with bug fixes, have been released in Algonomy CXP products in the release version 25.14.
Find
New configuration interface for Golden Enrichment Mappings in Find
A new configuration feature has been added to Find to support the effective management of Golden Enrichment Mappings. These curated keyword-value pairs serve as strategic inputs to the AI, helping it enrich thousands of queries by providing essential context, particularly for zero-result and low-performing searches.
Accessible through a dedicated tab in the LLM settings, the interface offers a clean tabular layout where users can view, add, edit, or remove mappings. It includes validation to prevent duplicate entries and sorts keywords alphabetically for ease of use. The configuration is designed to focus on a concise set of impactful mappings that deliver the greatest value to AI-driven enrichment.
Jira: ENG-30298
Channel-level insights added to Zero Search Results report
The Zero Search Results report in Find now includes a new Channel dimension, enabling merchandisers to identify which channels are contributing most to zero-result search terms. This enhancement supports both aggregated and de-aggregated views, helping teams uncover patterns across different touchpoints.
Users can filter the report by channel in both the graph and table views. The new dimension works across all regions and search term categories, including overall terms as well as those with and without facets, providing a comprehensive view of performance gaps.
Jira: ENG-30471
Social Proof
Enhanced campaign visibility in Social Proof list view
The Social Proof experience list view now includes key campaign details to support better visibility and decision-making. Users can view the start date of each campaign, see whether any variations are enabled, and identify the importance level assigned to each experience—all directly within the list view.
These enhancements help merchandisers quickly assess the status and relevance of multiple experiences, making it easier to prioritize and manage active campaigns efficiently.
Jira: ENG-30426
Recommend
AI-powered chatbot launched for help documentation
An AI-powered chatbot is now available in the Algonomy dashboard to assist users with platform-related queries. Accessible via a floating icon on the dashboard home and Recommend pages, the chatbot offers quick answers to common questions about setup, strategy usage, product features, and best practices. Users can either type their queries or choose from suggested prompts, and receive clear, structured responses with direct links to relevant documentation.
Key enhancements include:
-
Defined prompts and structured replies to ensure clarity and consistency in responses.
-
Relevance-based source ranking, showing the top three most useful documentation links.
-
Expanded content coverage including the Recommend section, Release Notes, Strategies Library, and 2025 summaries.
-
Live webpage integration for up-to-date documentation access.
-
Milvus-based vector search for faster vector search performance.
-
LangGraph-powered agent setup to manage documentation queries efficiently.
Note: This feature is currently available to Algonomy employees and will be rolled out more broadly in future phases.
Jira: ENG-29912, ENG-30340, ENG-30457, ENG-30458, ENG-30008, ENG-30009, ENG-30010, ENG-30351, ENG-30517, ENG-30518, ENG-30432
Platform
Support added for primary category based on custom attribute for streaming clients
Streaming catalog clients can now assign a product’s primary category based on a custom attribute, aligning with existing support available for feed-based integrations. This enhancement allows more flexible and consistent category management using client-defined attribute values during catalog updates.
Jira: PLAT-4053
Feed-to-streaming conversion support for catalog updates
A new capability has been added to convert catalog feed files into streaming format and route them through the streaming pipeline. This allows feed-based customers to leverage Find features and benefit from near real-time catalog updates.
The process supports both full and delta feed files, and includes handling for language, region, and SKU overrides. This enhancement bridges the gap between feed and streaming systems, offering faster data availability and improved flexibility.
Jira: PLAT-3996
Other Feature Enhancements
The following feature enhancements and upgrades have been made in the release version 25.14.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Recommend: Improved /placementsList API for faster access to placement profiles |
The /placementsList API has been enhanced to support quicker and more targeted retrieval of Recommend placement data. These updates allow users to view a streamlined list of Recommend pages and easily access placement profile details, including associated rules and performance metrics. New filtering options have been added to limit the response to Recommend placements only, exclude layout and QA configuration data, and filter by page type. These changes help reduce response size and improve load times, making it easier to manage and analyze placement performance. |
10-Jul-25 |
|
|
Recommend: Glass Views option in DNR rules now includes configuration check |
A validation has been added to the Glass Views option in Do Not Recommend rules. When users select the option to filter products viewed but not clicked in email placements, the system now checks if the required site configuration is enabled. If not, a warning message is displayed to guide users to enable recommendation view tracking through their Algonomy representative. This helps ensure the rule functions as intended and avoids misconfiguration. |
10-Jul-25 |
|
|
Ensemble AI: Click tracking and layout updates for Ensemble AI add-to-cart interactions |
Click tracking is now enabled for all products within an ensemble when users add them to the cart using the “Add to Cart” option. This allows each product added through an ensemble to be tracked individually, improving attribution and performance analysis for Ensemble AI. |
10-Jul-25 |
|
|
Find: Configurable UER handling for vector results |
Enhancements have been made to how the UER score is applied to vector-based results. The system now supports two options: rescaling the UER value within a configurable 0–0.1 range or removing it entirely from vector results. This provides better control over scoring behavior for improved relevance tuning. |
10-Jul-25 |
|
|
Find: Performance improvements for B2B parsers in Find |
B2B plugin performance has been optimized following detailed profiling and analysis. These enhancements address slowdowns related to secondary sort operations, improving response times without affecting existing functionality. |
10-Jul-25 |
|
|
Find: Support added for replica-type shard preferences in Find |
Find now supports shard-level read preferences based on replica types using Solr’s shards.preference parameter. With this enhancement, sites can configure preferred replica types (such as TLOG and PULL) for catalog collections, allowing more efficient distribution of search queries across available replicas. This update improves performance and flexibility by enabling non-leader replicas to serve read requests when configured. |
10-Jul-25 |
|
|
Find: Support added for reformulateQueryFields in Query Understanding config |
A new reformulateQueryFields configuration has been added to the Query Understanding setup for complementary search. This allows only the specified fields to be used for reformulating queries, instead of using all query tagging properties by default. |
10-Jul-25 |
|
|
Find: Hybrid search now bypassed for numeric queries |
A new numericAllowed flag has been introduced in the hybrid search configuration. When set to false (default), hybrid search will be skipped for numeric queries, helping improve result relevance and reduce unnecessary processing in such cases. |
10-Jul-25 |
Bug and Support Fixes
The following issues have been fixed in the release version 25.14.
|
Jira # |
Module/Title |
Summary |
General Availability |
|---|---|---|---|
|
Chatbot: Fixed chatbot icon visibility and print layout issues |
We have fixed an issue where the chatbot icon was not appearing immediately after login on the home page. Additionally, the layout issue causing oversized icons when printing the chat has been resolved. |
10-Jul-25 |
|
|
Chatbot: Fixed chatbot overlapping the Save button on key pages
|
We have fixed an issue where the chatbot icon was overlapping the Save button on pages like Advanced Merchandising and Configurable Strategies. The chatbot now avoids blocking interactive elements, improving usability across the dashboard. |
10-Jul-25 |
|
|
Chatbot: Fixed chatbot responses for greetings, exit, and unrecognized input |
We have fixed issues where the chatbot did not respond to greetings like "Hi" or "Hello", or properly handle chat exits and unclear inputs. It now responds with appropriate greetings, closes the chat with a goodbye message, and shows a fallback message when it doesn't understand the input. |
10-Jul-25 |
|
|
Fixed package boosting issue in new Find stack |
We have resolved an issue where package boosting was not functioning correctly in the new Find stack using Solr 9. Boosting is now applied as expected, ensuring accurate prioritization of products in search results. |
10-Jul-25 |
|
|
Fixed Solr server availability error in Find search |
We have resolved an issue that caused a "No live SolrServers available" error during product search requests for a client. The Find service now handles requests reliably without encountering Solr availability exceptions. |
10-Jul-25 |
|
|
Fixed file upload error in Data Science Workbench |
We have resolved an issue where file uploads appeared successful but resulted in FTP-related errors due to schema setup inconsistencies between Hive and Impala. File uploads now process correctly without triggering misleading error messages. |
10-Jul-25 |
|
|
Fixed issue with calculated cart states |
We have fixed an issue where the current cart state included items from multiple cart view events. The system now correctly returns only the items from the latest cart view as the active cart state. |
10-Jul-25 |