GuidesChangelogsAPI ReferencePublic RoadmapService Status
Changelogs
Back to All

March 2025

about 1 month ago by Hicham TAHIRI

Version 3.16.0

Released on March the 20th

Conversations

You can now search conversations by keywords, making it easier to find specific discussions.
This has been a highly requested feature, and we’re thrilled to finally bring it to you!


UI

In the bot list, we added a new column to easily identify which AI assistant is associated with each bot.



Version 3.15.2

Released on March the 19th

Knowledge bases

Improved visibility on failed source indexing: The knowledge base list now displays the number of successfully indexed sources (e.g., 82/83), making it easy to identify missing files.

Additionally, inside a knowledge base, unsuccessfully indexed sources are highlighted in red for quick identification and review.


The Knowledge Base window is now larger for easier navigation through multiple sources, and the 'View Content' button is more prominent for better accessibility.



Version 3.15.1

Released on March the 14th

Integrations

  • Tutogenius: A token was added to secure this integration

UI

  • UI improvements

Version 3.15.0

Released on March the 12th

New Models

Deepseek, Lite LLM, LLMProxy: This month, we're bringing you tons of new models to experiment with!



  • Lite LLM
    • Integration with LiteLLM, providing access to 100+ AI models through a unified API.
    • LiteLLM Documentation
  • LLMProxy
    • A LiteLLM fork used by Orange, optimized for secure and private model deployments within the Orange Group.
  • Deepseek
    • deepseek-chat: A powerful conversational AI model designed for natural and accurate interactions.
    • deepseek-reasoner: A model specialized in logical reasoning and AI-assisted decision-making.
    • To get started, get your API key here.
  • Anthropic
    • Claude 3.7 Sonnet: The latest model from Anthropic, offering enhanced reasoning capabilities and improved accuracy in complex tasks.
  • Google AI
    • Gemini 2 Flash: A high-performance model optimized for speed and real-time responses.
    • Gemini 2 Flash Lite: A lightweight version of Gemini 2 Flash, designed for faster processing and lower resource consumption.
    • Gemini 2 Pro Exp: An advanced model with extended capabilities for complex AI tasks and in-depth analysis.
    • Gemini 2 Flash Thinking: A variant optimized for enhanced reasoning and structured problem-solving.
  • Mistral
    • Mistral Saba: A model designed for deep comprehension and precise execution of complex instructions.
  • OpenAI and Azure OpenAI
    • GPT-4.5 Preview: The latest iteration of OpenAI's GPT series, providing improved accuracy and better contextual understanding.
    • o1: A versatile model focused on natural communication and adaptability.
    • o3-mini: A high-performance, compact model suitable for real-time applications with reduced resource consumption.

Assistants – Additional Filters

  • New filters to better manage AI assistants:
    • By bot
    • By knowledge base
    • By function
    • By AI deployment type

Knowledge Bases

  • New filtering options:
    • By bot
    • By AI deployment

Function Calls

  • New filters to better organize and track function calls:
    • By bot
    • By assistant

Conversations

  • Fixed the output counter in the Observability view.
  • Outputs are now displayed above instructions for better readability.

Knowledge Bases

  • A new colum shows the indexation success rate &number of sources succesfully / total number of sources in the knowledge bases reflecting the quality of the text extraction in the provided sources.

  • We now give more visibility on the failed indexed sources so that you can fix them.



Version 3.14.12

Released on March 6th

Fixes

  • Fixed an issue affecting the download of Revised Answers knowledge bases.
  • Improved knowledge base search: You can now edit search results directly.

UI Improvements

  • Standardized margins in pop-up windows for a cleaner interface.

Bot Creation

  • You can now choose between two bot types when creating a new bot:
    • Classic bot (NLU only) → Redirected to the Builder.
    • Advanced bot (AI Agent + NLU) → An assistant is automatically created and linked to the new bot, then redirected to the assistant configuration page.


Version 3.14.11

Released on March 4th

Integrations

  • Deploy your bots directly to Microsoft Teams! 🎉


Evaluations

  • Added a download button for the latest evaluation results.

Version 3.14.10

Released on March 3rd

DevOps

  • Improved load management on SNP 1.06 NLU.

Evaluations

  • UI improvements in Evaluations.
  • Explainations now appears in the results table to give you more details on each test result.

Conversations

  • Better display for Arabic messages