Release Notes for RapidCanvas Feb 27, 2025

New Features

Following are the new features introduced in this release:

Nomenclature Changes

To enhance clarity and consistency across the platform, we have made the following updates:

  • Prediction run is now referred to as Predictions.

  • API Connector Recipe has been renamed to Code Recipe.

Prediction Scheduler Functionality

We are excited to introduce the Prediction Scheduler within the Predictions component of the project canvas. This feature allows users to automate prediction jobs at scheduled intervals across four problem types: binary classification, binary experimental, regression, and multiclass classification.

Key Highlights:

  • Automated Data Pre-processing: The scheduler ensures data refinement before feeding data into the trained model, executing pre-processing and feature engineering on uploaded datasets.

  • Automatic Updates: If a dataset is regularly updated at the source, the pipeline runs on the new dataset without requiring manual uploads.

  • Execution Logs: Users can view logs for any errors encountered during prediction jobs.

  • Download Options: After each prediction run, the output dataset can be downloaded or added to the canvas.

  • Project Canvas View: Users can view the project canvas as it appeared at the time of the scheduled run.

  • Manual Trigger: Users can manually run a scheduled job at any time.

Edit Option for DataApp Chart Outputs

The Edit feature for DataApp chart outputs now offers users greater flexibility for visualization customizations.

Customization Capabilities:

  • Change chart color, type, and title.

  • Modify charts directly through a query box by typing desired changes.

  • Once saved, the original chart cannot be reverted, ensuring a tailored and consistent visual experience.

Publish Model Pipeline Updates to Prediction Scheduler

A new Publish to Prediction Scheduler option has been added under the Actions menu on the canvas, which allows users to publish updates from the model pipeline directly to the Prediction Scheduler.

Key Enhancements:

  • Republish Option: Fetch updates from the model pipeline in a specific scheduler.

  • Compare Changes: Compare the current pipeline with the scheduler version to track differences.

Global Search Functionality

A Global Search feature has been introduced, allowing users to search for various elements across the platform from a unified search bar available at the top of each page.

Searchable Elements:

  • Projects, Recipes, Datasets, Charts, Models, Artifacts, DataApps, Environments, Connectors, Predictions, Schedulers, Vector Databases, and Files.

Search results are filtered to match keywords, enabling quick navigation directly to relevant pages.

Import Text Files Directly on Canvas

You can now import text files directly onto the Canvas, streamlining the process of working with text-based files.

Key Enhancements:

  • Supported Formats: .txt, .json, .html, and .md files can be uploaded directly.

  • Recipe Execution: Only Code Recipes and supported Template Recipes can run on text files, keeping the original file format intact.

  • Contextual Actions: Right-click to access tabs for data contents and the source file, with options to update, execute recipes, or delete.

Support for RAG Recipe

The Retrieval-Augmented Generation (RAG) Recipe has been introduced, allowing users to convert data into vectors and store them in a vector database for AI-driven solutions.

Key Capabilities:

  • Generate and display vectors on the Canvas.

  • Use Code Recipes to generate or consume vectorized data.

Code Assist in Code Recipes

Code Assist is now available in Code Recipes, offering AI-powered coding support directly within the platform.

Key Features:

  • Code Generation: Enter a prompt to generate new code snippets or get explanations for existing code.

  • Multiple AI Models: Choose between GPT-4o, GPT-4-turbo, and GPT-3.5-turbo for optimal results.

  • Manual Code Copy: Generated code must be manually copied into the Code Recipe.

  • Clear History: Option to restart conversations, with automatic history clearance upon session expiry.

Right-Click Options on Canvas

Users can now access quick actions by right-clicking anywhere on the Canvas, offering options similar to those available via the plus button.

Right-click Options Include:

  • Dataset, Artifact, Model, Text File, and Recipe actions (AI-Assisted, Rapid Model, Template, and Code Recipes).

Libraries in Recipes Take Precedence Over Environment Libraries

Now, recipe-specific libraries take precedence over environment-level libraries, improving dependency management.

Key Details:

  • Custom libraries can be defined at the recipe level.

  • Isolated Environments: Recipes run in isolated environments with libraries specified at the recipe level, reducing version mismatch issues.

Enhancements

PostgreSQL as a Destination

PostgreSQL connectors can now be used as a destination for jobs and output datasets.

Key Highlights:

  • Expanded Storage Options: Users can now select PostgreSQL as a destination for saving data.

  • Append or Replace Data: Users can append new data to existing tables (if schemas match) or replace the dataset entirely.

Search Count in Create Connection

The matching connector count is now displayed when searching within the Create Connection window, streamlining the search process.

Key Highlights:

  • Search Efficiency: The number of matching connectors is now shown, making it easier to locate the right connector quickly.

Support for XLSX in Prediction Service

The Prediction Service now supports XLSX file formats, expanding the file types that can be used for generating predictions.

Key Highlights:

  • File Format Support: In addition to CSV, users can now upload and generate predictions from XLSX files.

  • Increased Flexibility: This removes the need for users to convert files into CSV format for prediction tasks.

Output Destination Node Visibility

The destination node for saving output datasets is now visibly displayed on the Canvas, improving navigation and tracking.

Key Highlights:

  • Visibility Improvement: The output destination node is clearly visible on the Canvas, ensuring easy identification.

  • Search Functionality: Users can quickly locate the destination node using the search feature.

Manual Prediction & Prediction Scheduler via Right-Click

Users can now access both Manual Prediction and Prediction Scheduler directly by right-clicking on a Model icon on the Canvas.

Key Highlights:

  • Quick Access: Right-clicking on a Model icon enables easy access to both manual prediction and scheduler options.

  • Improved Workflow: Simplifies the process of running one-time predictions or scheduling automated prediction jobs.

Python Version Selection for Streamlit Apps

When importing a Streamlit DataApp, users can now choose the Python version to ensure compatibility.

Key Highlights:

  • Version Compatibility: Python 3.8 and 3.10 are supported for Streamlit-based DataApps, ensuring smooth integration with the platform.

Starter Prompts in AskAI for DataApps

Users can now add up to 10 custom starter prompts in the AskAI window for DataApps, helping guide users in interacting with AI-powered DataApps.

Key Highlights:

  • Custom Prompts: Add predefined queries to assist users in getting started with DataApps.

  • User Convenience: Especially helpful for business users who need relevant starting points for their queries.

System Context in Prediction Service DataApp

A System Context field has been added to Prediction Service DataApps, allowing for custom AI-generated responses based on predefined prompts.

Key Highlights:

  • Custom AI Responses: Users can enter system prompts to generate consistent and contextually relevant AI responses.

Run Time Display for Recipes

The last successful run time of a recipe is now displayed for better tracking of recipe execution.

Key Highlights:

  • Tracking Improvement: Easily monitor the last successful execution of a recipe, aiding in workflow management and transparency.

Python Library Tab Updates

The Python Library Tab on Environment pages has been updated for enhanced clarity.

Key Highlights:

  • Clearer Information: New information icons have been added to provide better context and explanations regarding Python libraries.

Text-Based Pipelines Supported by Scheduler

Text-based pipelines can now be scheduled, increasing the versatility of the pipeline creation process.

Key Highlights:

  • Expanded Scheduling: Users can now create schedulers for pipelines that utilize text as both input and output, broadening the scope of data processing workflows.

Restricted Permissions for DataApp View Role

Users with the DataApp view role now have restricted access to editing, deletion, and configuration settings.

Key Highlights:

  • Role Restrictions: Users with this role will not have access to modify, delete, or configure certain features, ensuring more controlled access.

Adjustable Column Widths in DataApps

Users can now adjust the column widths within a DataApp, improving the data viewing experience.

Key Highlights:

  • Customization: The column widths can be customized to suit individual preferences, providing more control over data presentation.

These enhancements are designed to improve flexibility, usability, and control across various features of the platform.