Release version 7.4.1

Release version 7.4.1 includes new features, improvements, and bug fixes.

Unstructured-2-Structured - Beta: Introduced a feature to extract data from PDFs into a structured lakehouse. This update allows configurable field extraction, modeling data as tables for advanced and analytical querying. The phased release starts with Fat Table extraction and ingestion, followed by star schema support. The user-friendly interface ensures efficient data handling and high performance, enhancing accuracy and customer satisfaction. We have successfully implemented a proof of concept (POC) for Medicare.
Bing Search Fallback for Responses: Bing Search has been integrated as a fallback mechanism for Skypoint AI Copilot. This enhancement enables the Copilot to retrieve relevant or additional information when internal data sources are insufficient, improving the robustness and accuracy of responses and enhancing user engagement and satisfaction.
Improving Sherloq - Reflection Support: This feature enables models to learn from past performance, ensuring consistent error handling and better problem-solving abilities. This will enable copilot to learn from past experience (manually injected) thereby improving thinking process, forming highly accurate queries to data sources over time.

Enable Domain Grounding on Skypoint AI Studio: Introducing the Domain Grounding feature for the private copilot in Skypoint AI Studio. Users can now enable or disable Domain Grounding, providing 'Domain Context' and 'Terminology' for better domain-specific customization. This enhances the copilot's accuracy and relevance in your specific field.
MRI PMX Connector: This API-based connector will seamlessly complement user BAK backup file, providing nightly updates to maintain 'live' data within the Skypoint AI platform.
PayNW Connector Enhancement: Enhanced the connector for importing employee timesheet data onto the platform, streamlining data integration and improving overall efficiency.

AI Bot for Teams: We're excited to announce the upcoming Skypoint AI Copilot on Microsoft Teams! This integration provides quick and convenient access to Skypoint AI Copilot within the Teams environment, enabling users to chat with their data anytime, anywhere, enhancing productivity and improving the user experience.
Pre-defined Questions on Landing Page and New Chat: In an upcoming release, users will see a welcome message and suggested questions when they land on the Skypoint Copilot page or start a new conversation. Additionally, admin users will have the capability to set up these suggested questions through the AI Tuning module in Skypoint AI Studio. This feature aims to streamline user interactions, providing a more guided and efficient experience right from the start.
Quick Retry for System Failures: Skypoint AI Copilot users will soon be able to instantly retry prompts with a simple button whenever system failures occur, ensuring uninterrupted workflow. This enhancement minimizes downtime, ensuring smooth operation and improving reliability and user experience.

Enhanced Structured Data Configuration: New features will soon allow users to seamlessly configure end-to-end structured data on Skypoint AI Studio. This includes facilitating data annotation, setting up the data model context, and providing example SQL queries. These improvements are designed to enhance the accuracy and usability of Skypoint AI Copilot, ensuring a streamlined and efficient user experience.
Copilot UX/UI Improvement: In the future release, we're improving the Skypoint AI Copilot with a focus on user experience. Expect a simplified interface, smoother interactions, and efficient conversation history search. These updates aim to deliver a seamless, enjoyable user experience while maintaining high standards of performance and accessibility.
Instructions Tuning via Studio: Enhance AI performance by tuning instructions within Skypoint AI Studio. Users can specify domain information, provide details about data sources, and input prompt examples to refine AI reasoning capabilities, tailoring AI responses to specific user requirements.