Skip to main content

v1.1.2

Faster ingestion and answers, stronger data protection

  • Much faster document ingestion - Navigator now reads text pages with a fast text extractor and runs the heavy layout model only on pages that contain tables. A text-heavy protocol ingests several times faster (up to about 11 times faster on a mostly-text document); table pages keep the same accuracy.

  • Live page-by-page upload progress - uploading a document, creating a collection, and amending a document now show real per-page progress and elapsed time, instead of a spinner that looks stuck on a large PDF.

  • Faster start and lists - Navigator starts quicker after an idle period, and the study collection list loads faster.

  • More reliable chat - an answer now appears in your conversation even if the live connection drops during the reply. A rare case where the assistant finished with no visible answer now recovers the answer or explains a provider outage.

  • Type-to-search the study collection switcher - start typing to filter the collection list instead of scrolling.

  • Correct Insights totals - Total Queries, Time Saved, and Self-Service rate now match the per-site numbers shown in the same view.

  • Stronger protection of patient information - the step that removes patient identifiers before any external AI call is now reliable under load, so a busy moment no longer blocks an answer.

  • Security fixes - corrected file-path handling in page serving, sign-in rate limiting, sponsor data scoping, and chat and retrieval that stay strictly within the selected study.