Pragatix 2.14.3 - September 25, 2025
- 1 Main changes
- 1.1 General
- 1.2 Core Features
- 1.3 Performance & Usability
- 1.4 Reporting & Analytics
- 1.5 General Chat
- 2 Lite mode
- 3 Knowledge Assistant
- 4 Account Settings
- 5 Account Firewall Auditing
- 6 Guardrail Policies
- 7 Account Management
- 8 Account Reports > Firewall report
- 9 Account Reports > Content report
- 10 Site Managemen > Site Statistics
- 11 Small improvements
Main changes
General
Product Renaming: BusinessGPT is now Pragatix—the new name is consistently reflected across the entire product.
Core Features
Create chats from multiple files and folders within the Company Knowledge Base.
Add files to multiple collections, making organization more flexible.
Redesigned flow for creating a Knowledge Assistant with streamlined setup for sources and permissions.
Support for adding files and folders to multiple Knowledge Assistants.
Support for updating existing Knowledge Assistants with new files and folders.
Built-in safeguards to prevent granting permissions beyond those allowed by the source.
Add to Favorites Option:
Users can now mark Assistant chats and Data Analysis chats as Favorites, making them quicker and easier to access.
Performance & Usability
Custom Branding Simplified: Predefined configuration options for logos, background images, colors, and main text, plus support for uploading custom CSS files.
Reporting & Analytics
Firewall Reports: Now filterable by time, users, and groups for better monitoring of adoption and risk.
Interactive Reports: All account reports are now clickable—select any data point to view detailed information.
New Charts & Filters:
AI Service by Time: Visualize usage trends of AI services.
Questions by Time: Track question activity over custom time periods.
Time-based filtering has been added to Account Content Reports.
General Chat
Prompt Library: Added a centralized library of prompts to streamline and enhance general chat usage.
Interrupt Responses: You can now stop AI responses mid-generation, giving you instant control and reducing wait times.
Enhanced Source info: Document preview, sensitivity source, and type are presented in the response.
Lite mode
After receiving an answer in the chatbot, users will now see:
Sensitive Document Notification – A banner highlights when a document is classified as highly sensitive, ensuring better visibility and awareness.
Source Icons – Icons (SP, Document, Site) make it easier to identify the document’s source.
File Type Icons – Clear visual markers for Word, PDF, and Excel files for faster recognition.
Sensitivity Level Indicators – A color-coded circle showing the sensitivity level of each document for quick confidentiality awareness.
Users can start chats from multiple files and folders in the Company Knowledge Base.
Response Control: Users can now interrupt AI responses as they’re being generated, saving time and giving instant control.
Favorite Chats: Users can now save Assistant and Data Analysis chats as favorites for faster and easier access.
Added a Max Words in Answer option under advanced chat settings, allowing control over response length.
Added the ability to delete individual chats from chat history.
Knowledge Assistant
We’re excited to introduce a new Knowledge Assistant page, designed to make setup and management simpler and more powerful:
Redesigned Creation Flow – A streamlined process for setting up Knowledge Assistants, with an improved flow for configuring sources and permissions.
Multi-Assistant Support – Add files and folders to multiple Knowledge Assistants with ease.
Update Existing Assistants – Seamlessly update Knowledge Assistants by attaching new files and folders.
Built-in Safeguards – Enhanced security prevents granting permissions beyond what is allowed by the source.
Account Settings
Simplified Custom Branding: Easily configure logos, background images, colors, and main text, with added support for uploading custom CSS files.
Account Firewall Auditing
Added a records selector, allowing users to adjust the number of entries displayed in lists for easier data review.
Added two new components for the algorithm filter:
Selected Documents from Search – Identifies complete documents from relevant chunks within a large content scope that can be fully included within the effective context limit.
Dominant Document from Search – If all documents cannot be included, the system selects a dominant document, defined as one with at least 60% of its content returned in the search results.
User names from embedded chats now appear as Anonymous in auditing.
Guardrail Policies
Added an option to hide disabled policies.
Account Management
Split Instructions into two separate pages: Response Guidelines and Startup Prompts, improving clarity and organization.
A new Prompt Mode field is now available when creating prompts in the Propmt Library.
Account Reports > Firewall report
Added a Questions by Time chart, providing a visual overview of question activity trends over selected time periods.
Added an AI Service by Time chart, enabling visualization of AI service usage trends over time.
Account Reports > Content report
Added a time period selector for account usage, allowing users to view data by month, last 30 days, last 3 or 6 months, or all time.
Added From/To date range pickers, allowing users to filter data by custom time periods for more flexible analysis.
Site Managemen > Site Statistics
New button allows quick removal of users or accounts.
Small improvements
Added support for searching specific URLs within a site.
Reduced the size of headings in answers for a cleaner and more consistent display.
Made all charts in the Site Firewall interactive— clicking any widget now opens the underlying events.