This release introduces significant advancements leveraging generative AI. Key improvements include an enhanced embedding model with multilingual and cross-lingual capabilities, a smarter PDF parser that utilizes contextual classification, and new reasoning models for deeper analysis. Additionally, this version integrates with Azure AI Foundry, providing users with access to a wide selection of LLMs and VLMs.
🎯 Highlights
Enhanced AI capabilities
Multilingual Understanding
Intelligent Document Processing
Deeper Insights
Flexible LLM/VLM Access
Improved control over Scopes and Analyses
🧠 CodeBeamer Add-on Improvements
Enhanced Embedding Model with Multilingual & Cross-Lingual Capabilities
New state-of-the-art embedding model leveraging generative AI to deliver better performance
Superior performance on complex data, improving semantic representation even in nuanced cases.
Multilingual support, enabling understanding across a wide variety of languages.
Cross-language semantic search, allowing you to find related concepts across languages.
Richer, more detailed embeddings, leading to better document clustering, search accuracy, and knowledge representation..
New Reasoning Models for Deeper Analysis
Introduction of advanced reasoning models to handle more complex analytical tasks.
While these models may be slower, they provide significantly more detailed and nuanced analysis and an improved understanding of complex interdependencies.
Greater attention to edge cases, ensuring better support for specialized or multi-layered queries.
Improved Searching, Scoping, and Analysing
Possibility to configure the similarity threshold both in Requirement Discovery and Search
Addition of report-based scopes in the Requirement Discovery
Introduction of a scope selector in the search functionality to enable more fine-grained control over your search results
Possibility to separate the analysis into item level and project level
User experience improvements, such as the highlighting of linked items and an enhanced visualisation of related items
Azure AI Foundry Integration – Choose Your LLM
Azure AI Foundry integration, giving users access to a wide selection of paid and open-source LLMs and VLMs, including: GPT, Claude, Mistral, Grok, and many more.
Flexible inference options based on task complexity and performance needs.
Trade-offs in latency and cost, with higher power models offering richer outputs at increased price/token and slower response times.
📄 PDF Importer Improvements
Smarter PDF Parser with Contextual Classification
Introduction of generative models for contextual understanding of item hierarchy
Improved classification of headings and textual fields, now using surrounding context for greater accuracy.
Better extraction of hierarchical structures, capturing nested relationships in complex documents.
Higher-quality and more flexible item splits for structured data like requirements, specs, and procedural documents
📋 Technical Notes
System Requirements
No changes to existing system requirements
All improvements are backward compatible
Known Limitations
Model availability is continuously updated
Different model versions may have varying performance characteristics
Configuration Notes
Users can now select from multiple AI models based on their specific needs