Skip to main content

GPTfy 1.13 Release Notes

Release Date: Jan 10, 2026

New Features​

  • Salesforce Data Cloud Integration : Extended Data Context Mapping to work with Salesforce Data Cloud objects, supporting data from both local and cross-organization Data Cloud instances.
  • File-Based RAG Sync : Introduced file-based RAG Sync, allowing users to upload files directly into vector stores for agents to use as context for knowledge grounding.
  • Agent File/Image Attachment : Users can now upload images and files directly within agentic conversations via a new link icon in the chat window, enabling agents to respond based on the uploaded content.
  • Agentic Build UX : A new interface allows for intuitive creation and management of Agents. Agents can be described in a dedicated field, which auto-fills the System Prompt and suggested skills. RAG Sync records can be attached under Knowledge Base when needed.
  • Agentic Testing Framework : Run large-scale tests for agentic and standalone prompts with Test Cycles. A unified batch process manages all inputs under one secure ThreadId, and results can be auto-verified with an AI accuracy check.
  • Trending Reports : Introduced six new trend analytics reports to visualize usage across Object, User, Prompt, Profile, Role, and Department, accessible via a dedicated Trending Cockpit card.
  • Cockpit UX Redesign : The enhanced Cockpit UI offers a modern, streamlined experience with three dynamic floating tiles for real-time tracking of Amount Saved, Time Saved, and Active Prompts. The layout includes structured tabs for Analytics, Agentic, and Advanced settings.
  • Prompt Builder UX Redesign : Introduced a simplified, intuitive Prompt Builder interface. Users can describe their prompt in a single input box or choose from four pre-loaded design specifications that auto-generate required text and context mapping. A "Build Your Own" option is retained for full customization of JSON-based, Canvas, and Agentic prompts.

Enhancements​

  • GPT-5 Model Support : Integrated support for the GPT-5 model, including configuration for low reasoning and updated maxOutput Token parameters.
  • Enhanced Cloning/Export Consistency : Improved the Clone with Related, Global Import/Export, and Prompt Version/Reinstatement processes for Agentic Prompts to consistently preserve the Agentic Function Class and relevant parameters (Prompt command, AI Model, Permissions, Profiles, Description).
  • Enhanced Security Audit Viewer : Improved the Security Audit Response Viewer to display chat-specific data when the audit record originates from a chatbot conversation.
  • Voice Model Flexibility : GPTfy now supports Connector Class & Processing Class for voice-based AI models, allowing users to integrate any AI model using these classes for greater voice interaction flexibility.
  • Agent Auto-Navigation : Added an optional feature to automatically navigate to newly created or updated records via Agents, enhancing workflow efficiency.
  • Improved Agentic Prompt Validation : Enhanced validation to ensure all essential fields are verified before activating Agentic-type prompts.
  • Agentic Prompt Page Enhancement : Improved the Agentic prompt layout by removing several fields and tabs (e.g., Test, Grounding, Response Validation, Actions) to streamline the user experience.
  • GPTfy Branding Updates : Refreshed Chatbot branding in the Agentic interface to match the updated GPTfy design system, including revised colors, icons, and backgrounds. The user avatar now displays the user's Salesforce profile picture.
  • Chat UI Improvement : Implemented a UI update to disable the chat input section until a valid response is received, improving user experience.
  • Azure Search Support in RAG Sync : Extended RAG Sync capabilities to support Azure Search, expanding storage and retrieval options.
  • Account Plan Support : Added support for Account Plan object, enabling users to leverage account planning data within GPTfy.

Issues Fixed​

  • Security : Implemented validation to prevent the creation of duplicate Security Layers with the same Apex class name and ensured Response Validation remains non-editable after prompt activation.
  • Prompt Actions : Resolved the issue where Prompt Actions were still triggered despite Response Validation failure; actions now execute only when validation succeeds.
  • Canvas Auditability : Resolved an issue by introducing a "Prompt Element Details" field in the Security Audit to preserve historical Canvas dimensions, coordinates, and configuration details for full auditability.
  • Cloning Consistency : Resolved the problem where Permission Set, Loopback, and Response Validation values were not preserved when using the Clone with Related functionality.
  • Global Export/Import : Resolved issues with Global Import/Export functionality related to the "file processing model" was not being selected and failures when used in combination with Response Validation, Loopback, and Agentic function class.
  • Console Load Time : Optimized data retrieval in the GPTfy User Console to initially pull only the count of Security Audit records, significantly reducing loading times.
  • Chat End Message : Fixed an issue in the GPTfy Agent to ensure the 'End Chat' message is distinctly labeled as 'Chat Ended,' differentiating it from 'Chat Expired' messages.
  • Agentic Chat Hyperlinks : Addressed the issue where hyperlinks in Agentic chats were only visible on hover.
  • Data Context Mapping Save : Fixed a save failure issue where users were unable to save data context mapping after unchecking selected fields using the search option.
  • Validation Error Messages : Fixed "Null Object Error" in Response Validation to display the correct message for invalid Apex class names. Also resolved the issue where a success message was incorrectly displayed on validation errors during GPTfy Flow execution.
  • External Data Extraction : Verified and confirmed that the External Object Extraction issue at the third-level text hierarchy for Standard users has been resolved.
  • Miscellaneous : Fixed issues with Missing Prompt Component Validation, Incorrect Console Error Message, AI Mass processing + Response API, and Prompt Action Framework + Response API.

Known Issues​

  • File Based RAG Sync Image Upload : File-based RAG sync is currently unable to upload images.
  • Agentic Function Class Activation : The agentic prompt is activated despite having a bad value in the Agentic Function Class field.
  • Prompt Export/Import Missing Data : Prompt Export/Import feature does not include permission sets and the Time Saved Coefficient during import.
  • Catalog Card Export : The Catalog card is not exporting with the prompt during global export.
  • Prompt Catalog Configuration : Suggested adding an option for configuring permission sets in the Prompt Catalog during configuration.
  • Cockpit UI Static Indicator : In the new Cockpit UI, the blue line indicating the current tab is static, not dynamic.
  • Console Error Message : GPTfy console displays incorrect error messages for prompts that have inaccessable dynamic formula fields added in the data context mapping.
  • Prompt Catalogue Card Metrics : Time Saved and Cost Saved values are not getting populated in the Prompt Catalogue Card.
  • RAG Sync Status Issue : RAG Sync status Halts in In-Progress when an invalid filter condition is present in the RAG Sync filter.
  • RAG Sync Filter Validation : Filter condition validation is not performed before saving the RAG Sync record, allowing invalid conditions to be saved.
  • AI Mass Processing Limitation : AI Mass Processing is unable to run when more than 7 prompts are added to it.