RAG-Engine 1.3 adds hierarchical indexing and streaming answers

AI · 5 min read

The hierarchical indexing feature groups documents and assets into topic trees so retrieval prioritizes close context before broader knowledge, which helps models produce concise, relevant answers for UI design and style-guide queries. The framework includes automatic topic clustering and rules-based grouping to respect access controls and asset ownership.

Streaming answers are now supported end-to-end, enabling assistants to start returning partial, incremental responses while retrieval continues in the background. For design reviewers, this means faster first-pass suggestions with subsequent refinements as deeper context becomes available.

RAG-Engine 1.3 also introduced connector templates for design asset stores and version control systems, making it easier to index component libraries and changelogs. Enterprise users lauded the focus on reproducibility and audit trails, which are essential for using RAG systems in regulated design workflows.