Skip to main content

Release v1.19.2

D
Written by Daniel Schindler

We’re excited to announce the release of Version 1.19.3, delivering critical security fixes, enhanced document processing capabilities, and expanded integration across the Certara.AI ecosystem. This update introduces Synapse integration across all applications, improved multi-embedder support for Composer workflows, and significant performance optimizations—reinforcing our commitment to security, scalability, and reliability.


Certara GPT (certara-gpt)

Features:

  • Enhanced create embeddings functionality for CoAuthor sources with chunk ID returns

  • Create custom indexes in the vector databases via the Certara API.

Fixes:

  • Corrected rawtext field handling when document sectioning is enabled

  • Preserved embedder order in transformer service

Performance:

  • Restored debug logging for external model providers

  • Updated pod naming conventions for better system organization


Cortex API & Data

Features:

  • Updated Clinical Trials index to include Results section data

  • Improved PubChem data source update reliability

Fixes:

  • Resolved tag concept import failures in PubChem updates


Synapse

Features:

  • Upgraded to version 2.4.2 with enhanced document processing capabilities

  • Improved Document ID usage and help text for better user guidance

  • Enhanced model and data fabric population when switching between applications

Fixes:

  • Resolved non-fatal JavaScript error related to “synapse-cur-file” element

Performance:

  • Added backup functionality for improved data protection


Vyasa Account, Axon, Layar, Signal

Features:

  • Integrated Synapse branding and dropdown menu updates across all applications

  • Enhanced embedder modal to display JobID error messages for better user feedback


Security Updates

  • Fixed account registration security vulnerability

  • Enhanced session key validation

  • Applied multiple CVE patches for Release 1.19


Known Issues

  • The Summarize functionality is currently not working as expected in Composer workflows for systems using the OSS-GPT model. All other configurations, including those running Llama and similar models, are functioning normally. A fix for OSS-GPT systems is in progress and will be included in an upcoming patch release.elease.

Did this answer your question?