Release Notes for Knowledge Enrichment deliver a concise overview of new features, resolved issues, and key updates—helping you stay informed and get the most out of your solution.
September 2026
New Features
Data Curation API
- Added an image pre-filter that identifies natural photographs unlikely to contain meaningful text and skips OCR for those images, reducing unnecessary OCR processing. Images that may contain text, or whose classification is uncertain, continue through OCR.
Hyland Automate Integration
- Hyland Automate now includes native support for the Knowledge Enrichment Connector. Workflow builders can enrich content directly within an Automate process without writing custom code or building external integrations.
Product Updates
Data Curation API
- Resolved an issue where POST /config/options/rules and PUT /config/options/defaults silently discarded unknown or misspelled fields and returned a success response. Both endpoints now return 400 Bad Request for unknown fields at the request’s top level, directly within config, and within entries in conditions. Valid requests and existing, stored configurations are unaffected.
Security
- Remediated security vulnerabilities across the Knowledge Enrichment platform. All dependencies and runtime components have been updated to address known CVEs identified during routine security scanning. No user action is required.
July 2026
New Features
Data Curation API
- Added optional AWS Textract table detection to the Data Curation pipeline. When enabled, table detection identifies and extracts structured tables from image-based documents, preserving row and column relationships. The feature is opt-in, so users that don't require table detection can leave it disabled and avoid additional processing costs.
- Introduced a dedicated management API for creating, updating, and resetting per-tenant pipeline configuration. The new /config/options API provides a full configuration lifecycle with multiple endpoints spanning defaults and rules management.
- Added a lightweight, CPU-based image pre-filter that automatically identifies natural photographs and skips OCR for those images. The filter uses a cascade of fast computer-vision signals to reduce unnecessary OCR invocations and improve overall pipeline throughput.
- Data Curation now persists document text renditions directly in the Content Lake when the Content Lake is being used. Consumers can retrieve a markdown rendition of document text through the existing Content Lake API.
Product Updates
Context Enrichment API
- Resolved a condition where document processing jobs could become stuck or experience significant delays under certain conditions.
- Resolved a condition where guardrail validations were not applied to context enrichment results in certain processing scenarios.
Security
- Remediated security vulnerabilities across the Knowledge Enrichment platform. All dependencies and runtime components have been updated to address known CVEs identified during routine security scanning. No user action is required.
March 2026
New Features
Data Curation API
- Added optional named entity recognition to PII workflows. When enabled with PII detection or masking, Data Curation can now identify and report or mask names and organizations.
Context Enrichment API
- Introduced instruction support to Context Enrichment actions. Consumers can now provide instructions to all Context Enrichment actions to help guide the output and improve output quality.
Product Updates
Data Curation API
- Improved PII detection and masking accuracy for Social Security numbers, credit card numbers, addresses, and ZIP codes. Enhancements include better support for card numbers of varying lengths, long ZIP codes, apartment numbers, and refined SSN detection to reduce overlaps and missed values.
- Improved extraction quality for documents with complex tables through an updated Document Filters integration.
Context Enrichment API
- Added support for .jpg files in processing workflows.
- Updated process request handling to support kSimilarMetadata (example-based metadata templates) as a list of dictionaries.
- Improved guardrail-related error handling with clearer, more specific messages when requests are blocked, including dedicated messaging for prompt injection scenarios.
- Improved file upload validation so empty file uploads now return a clearer error message.
December 2025
Product Updates
Context Enrichment API
- Added enhanced protection against malicious prompts. Improved safeguards now detect and block unsafe or malicious prompts during document processing, reducing the risk of harmful content being executed or returned.
- Improved document processing reliability and stability across workflows to reduce failures and improve result consistency.
- Clarified and simplified error messages, helping users and developers troubleshoot issues more quickly.
- Updated request payload structure for actions and object keys and introduced explicit payload versioning. To ensure compatibility with the updates, clients must review and update any integrations that rely on existing action definitions or object key structures.
October 2025
New Features
Added PII detection and masking. When enabled, detected PII is replaced with [PII], and the type and count of detected information is reported.
Product Updates
Knowledge Enrichment
- Introduced support for processing HL7 Medical Records files.
- Improved performance and stability, including under high load conditions.
Data Curation API
- Updated embedding model selection with a specified list of supported models. Each model includes recommended default and maximum chunk sizes.
- Provided ready-to-use C# and Python examples to simplify integration and usage of Data Curation features.
- Added an option to normalize dash-like characters into the ASCII minus sign for consistent and machine-friendly output.
Context Enrichment API
- Increased the maximum request body size to 160 KB to allow larger payloads.
- Added built-in AI guardrails to support safe interactions.
- Updated sample scripts to simplify setup and usage.
June 2025
This release enables organizations to enrich their data lakes and catalogs with meaningful metadata, summaries, relationships, and embeddings, making enterprise content more discoverable, analyzable, and usable across AI workflows, governance systems, and decision intelligence pipelines.
New Features
- Support for 600+ file types: Process and analyze diverse content formats including structured documents (PDF, Word, Excel), multimedia files (images, audio, video), code repositories, email archives, and legacy formats, ensuring comprehensive coverage of your organization's data ecosystem.
- Contextual metadata: Extract and create rich, semantically-meaningful metadata including topics, themes, entities, document types, and business context, transforming unstructured content into structured, query able information assets.
- Contextual text chunking: Intelligently segment documents into meaningful chunks while preserving structural context, section hierarchies, and positional references, enabling more accurate retrieval and maintaining document relationships for downstream AI applications.
- Image description: Automatically generate detailed, contextually aware descriptions of images, charts, diagrams, and visual content within documents, making visual information searchable and accessible for content discovery and compliance workflows.
- Embedding generation: Create high-dimensional vector representations of content that capture semantic meaning and relationships, enabling advanced similarity search, content clustering, and AI-powered recommendation systems across your knowledge base.