- Skypoint AI introduces significant advancements including enhanced Short-Term Memory for improved conversational coherence, superior tabular data extraction capabilities for accuracy in unstructured data analysis, and the option for longer conversation names for better organization.
- The platform now offers a Self-Service AI Configuration feature, allowing users to easily and intuitively configure unstructured data sources for AI model training, simplifying the process of enhancing AI capabilities.
- For Skypoint Copilot, We've upgraded our response formatting for even clearer and more structured answers to your unstructured prompts.
- Our Copilot has refreshed color palette which not only enhances the visual appeal but also provides a more comfortable and eye-friendly interface, making your interactions more enjoyable.
- Skypoint Copilot's latest update revolutionizes user engagement with direct feedback options, seamless conversation sharing, precise timestamped conversation histories, and enhanced interaction during wait times.
- Paragraph Formatting in Unstructured Responses .
- Embrace the power of our new Usage Report Implementation, a comprehensive system for in-depth monitoring and analysis of your structured and unstructured data usage.
- Unveiled the Skypoint CoPilot, a substantial upgrade aimed at refining user experience. This release features a more sophisticated response presentation, enhanced chat functionalities, and a comprehensive conversation history management system.
- Revamped 'New Support Request' Section: An upgrade to the 'New Support Request' section within the Studio App focuses on making the feature more intuitive and user-friendly, streamlining the process for our users.
- We've reengineered our system architecture using AstraTunnel, ensuring a more robust and efficient performance tailored to each client's unique needs.
- Each tenant now benefits from a dedicated Astra DB provisioning, offering enhanced data management and security.
- SQL Query Support has been provided for Skypoint AI responses for structured data. Structural Decomposition has been enabled for more context in Constructing SQL Queries. A tool has been built, which efficiently validate SQL queries from LLM.
- Skypoint AI has been added in the product listings dropdown of "Help + Support" Section .
- Chat model has been upgraded from 8k to 16k Token Limit to support more extensive reasoning traces.
- Citation Support - Enhances reference inclusion within unstructured responses, improving content reliability and research integrity.
- Optimized data model for select customers, thereby, improving the accuracy and performance of AI responses from structured sources.
- Enhanced AI Performance - Model tuning for selected customers' data improves overall system performance.
- Introduced support for data model context that enhances text2SQL precision.
- Automated provisioning of key spaces in Skypoint AI's enablement workflow.
- Using the MultiQuery retriever for unstructured querying, We can now produce a variety of queries based on different viewpoints for a single user input query. By creating diverse angles on the same query, we are able to surpass some constraints of distance-based retrieval, yielding a more comprehensive set of outcomes
- The ADLS Gen2 V2 connector has been improved to facilitate incremental data ingestion with CSV files.
- Enhancements to the user interface have been implemented in the Granular Scheduler, simplifying the configuration of Pipeline Schedules.
- When reviewing records within the data unification process, we now have the capability to view the details of both matched and unmatched records, not just their respective counts.
- The OS Langchain was forked, a toolkit developed to extract table schemas and annotations from Unity Catalog Tables using Databricks RestApi, then integrated into the Skypoint project, resulting in a substantial 40% improvement in structured data response precision.
- Users can now adjust email verification time and send DSR verification reminder emails in Skypoint Studio, with the default time previously set at 24 hours.
- Enabled the use of Service Principal for Databricks Lakehouse Provisioning.
- Enhancing coding standards and implementing changes to enhance efficiency.
- The Scheduler enhancements enable automated and orchestrated task execution in the pipeline.
- WelcomeHome connector activities load to be support incremental run
- DSR to support email for verification, and additional settings for expiry.
- Beta version of Skypoint AI plugin released
- Databricks runtime is upgraded to 12.2 LTS
- dbt Core is upgraded to version v1.5
- The Python version for all Azure functions is upgraded to version 3.10
- Previous 3 versions of map now available for users to compare
- Run history now displays Master Data type.
- Run history option is now available for timelines.
- Folder based entity ingestion using Azure Data Lake Storage Gen2 Connector
- DBT Transformations
- Type of refresh in resolve
- Rule Match accuracy improvements
- Higlight of Map attribute
- Usage based report
- MDM support for predictions
- Improvements in CLV model
- UKG import connector
- Warning to user in map screen when import is updated/
- Invitation email updates
- Displaying datatypes on map screen and enabling entity wise run
- SkyPointGPT can be enabled for usage on instance data
- WelcomeHome import connector
- External tables in Lakehouse accessible in all downstream processes/
- View different master data on Dashboard.
- Usage of the Lakehouse SQL-created tables.
- Deidentification of PII and PHI data.
- Added SkyPointGPT.
- Domain selection in instance.
- Allow the same priority for multiple attributes in ML match.
- Added exceptions in the ML match.
- Added Snowflake DLT integration.
- Added Type in master data.
- Autosuggestion in a SQL editor for Audience creation.
- Data is available to the user for newly ingested tables or inherited tables.
- Deletion of the instance/tenant will delete respective storage, Cosmos, Unity Catalog/Schema/Tables, and SQL warehouse.
- Create audience from the prediction models.
- Added topic modeling and filter on sentiment model view page.
- View multiple master profiles.
- Migration of hive_metastore to Unity Catalog.
- Merge execution per master data.
- Added Transformations.
- Multiple Master data creation per instance.
- Added RFM model view page.
- Added SQL query editor in Audience.
- Added DateTime option in Metrics.
- External table creation in Lakehouse with the user-defined name.
- RFM (Recency, Frequency, Monetary value) model is added under Predict > Built-In.
- Added search field in the dropdown to select the desired value.
- Added multiple timeline journeys per instance.
- Setup instructions for import connectors.
- Added profile filter ("Email approachable" or "Phone approachable") to create a quick audience.
- Real-time status update on Audience page.
- Usage of tables across all downstream processes.
- Real-time status updates for Exports, Audience, and Prediction models.
- Bulk Processing of Data Subject Request (DSR).
- Data ingestion through Delta Live Tables (DLT).
- Convert Pandas data frame into Spark data frames.
- Import connectors redesign.
- Segregating data in Bronze, Silver, and Gold.
- Prediction model is migrated from Azure ML to Databricks using ML flow.
- Net Promoter Score (NPS) calculation in Business Metrics.
- Auto provisioning of Lakehouse SQL during the instance creation.
- Run history shows all stages of ingestion.
- Removal of CDM dependency for all downstream processes to improve pipeline performance and stability.
- View profile card from Audience.
- Timeline creation from Audience.
- The storage layer is restructured and organized into three layers named Bronze, Silver, and Gold.
- Replaced the Standard cluster with an Automated cluster to support optimized autoscaling.
- Added Refresh personal access token in SQL Access to configure the expiry time for an access token.
- Support Change Data Capture (CDC) in Lakehouse SQL.
- Added reason for the discarded entities in map tables under lakehouse.
- Added access token to integrate with Visualization tools through Lakehouse SQL.
- Refresh Hive metastore is added for syncing new entities in Lakehouse SQL.
- Skypoint Lakehouse made available for bidirectional integration with third party tools like Fivetran, Dbt, Power BI, Tableau, etc.