CoAuthor
What's New
Create Documents from Your Own Templates
You can now start a new CoAuthor document directly from templates you've saved in your personal template library.
Automatic List of Abbreviations
CoAuthor can now automatically create and maintain a List of Abbreviations (LOA), keeping entries consistently formatted and alphabetically ordered.
Choose Your AI Model
Users can select which AI model CoAuthor uses for content generation. Administrators can also set a default model for their organization.
Improved Source Traceability in Word
Generated content can be traced back to its source material through Word comments. These traceability comments are now automatically hidden when content is inserted, keeping documents cleaner and easier to read.
Copy Entire Projects
Duplicate a project in a single step, including its structure and content.
Bulk Prompt Updates
Use Find & Replace to update prompts across a document. You can choose to search within prompt content or titles only.
Enhanced Narrative Previews
Preview entire subject narratives at once and view sections in landscape orientation for easier review.
Better Narrative Filtering
Filter for blank values and search directly within filter options to find information faster.
Built-In System Health Check
A new health-check feature helps monitor CoAuthor system reliability.
Improvements
Faster loading for Data Sections → Fields.
Data Sections now remember your scroll position while navigating.
Unsaved-change warnings have been expanded to help prevent accidental data loss.
Search now includes content stored in child folders.
Sources can be selected at the document level without opening individual documents.
Improved background job processing reliability.
Prompt libraries now sort numerically (for example, 1.1, 2.1, 10.2).
Simplified interfaces by removing unused actions and options.
Improved dropdown sizing and macOS scrollbar styling.
Audit logs now clearly indicate that timestamps are displayed in UTC.
Informational table headers are preserved in generated output.
Bug Fixes
Document Formatting
Resolved several formatting issues in generated content, including:
Font colors and highlighting
Bold table headers
Multi-level bullet and numbered lists
Formatting inheritance from the Word insertion point
Also fixed spacing issues after line breaks and formatting loss in certain list items inserted through AI Chat.
Add-In & Document Management
Existing prompts now display correctly in the Word add-in.
Sources visible in the portal now appear correctly in the add-in.
Documents no longer fail to load when token expiration or embedding checks encounter issues.
Narratives
Fixed intermittent issues where subject lists disappeared or failed to refresh when filters changed.
Internationalization & UI
Excel documents now correctly support Japanese characters.
Corrected prompt library labels and LOA modal status displays.
Fixed sorting issues for columns named "Order."
Generation & Sources
Fixed Generate All progress tracking issues.
Resolved a source attachment issue that could carry source links between projects after re-importing.
Security, Performance & Stability
Security Updates
Applied platform and dependency upgrades, including critical security fixes and broader vulnerability remediation.
Performance Improvements
Faster source and file upload processing
Improved authentication and transaction performance
Faster pagination
Better handling of large files and ZIP archives
Background source deletion support
Stability Enhancements
Consolidated Word insertion logic
Fixed an undo (Ctrl+Z) regression after AI Chat insertions
Composer
What's New
Adjustable Reasoning Levels
Choose reasoning settings for supported external AI providers. Claude 4 models now automatically use the appropriate thinking mode.
Improved Reasoning Settings
The Show Reasoning option has been fixed, and incompatible provider options are automatically disabled when necessary.
Select an Embedder
Choose which embedding model is used for RAG (Retrieval-Augmented Generation) requests.
Fixes
Resolved streamed-text display issues.
Improved visibility of streaming errors.
Fixed citation handling when RAG features are disabled.
Corrected behavior of the Skip RAG setting.
Fixed embedder selection behavior.
Improved document image rendering and preview reliability.
Layar
What's New
Upload Visibility
See exactly which files failed during upload.
Connector Improvements
Simplified JDBC connector setup.
Browse Document Sets
Open document sets to view the documents they contain.
Simplified Access
Layar IDs now provide a cleaner access experience.
PII Scanning
Added personally identifiable information (PII) scanning during source ingestion, legacy-source scanning, and reporting.
Fixes
Restored parent-document previews.
Improved scrolling behavior on large displays.
Fixed search errors during high-concurrency usage.
Disabled an unsafe JDBC driver upload endpoint.
Curate (Clinical Trial Data)
Fixes
Restored the Results First Post Date filter.
Fixed the Date Added filter.
Restored values for the Trial Data: Design Masking metadata filter.
Codex
Fixes
Sidebar filters now persist after refreshing the application.
Pipelines (formerly Synapse)
What's New
New Name
Synapse has been renamed to Pipelines throughout the application.
New Node Types
Added:
Map-reduce processing nodes
SQL nodes that expose available tables and columns
Expanded MapReduce Support
MapReduce workflows now support table-based data in addition to text.
Asynchronous Execution
Pipelines can now be executed asynchronously through a new API endpoint.
Layar ID Support
Saved pipelines can now be called using Layar IDs.
Certara GPT
Improvements
AWS Bedrock Integration
Added AWS Bedrock support for model hosting and future scalability, while also resolving Bedrock embedding issues.
Expanded Model Support
Updated the model server to support newer models, including Gemma 4 and Nemotron 3 Super.
Embedding Reliability
Improved job handling and recovery mechanisms, reduced false timeouts, and enhanced processing stability for large embedding jobs.
Authentication & Account Management
OAuth tokens are revoked on startup when appropriate.
Inactivity timeouts are enforced.
License refresh handling improved.
SSO users can no longer edit identity fields.
Inactive users are removed from portal user lists.