Changelog
Stay up-to-date with all the improvements and new features added to Davinci.
Version 3.2.0 - Markdown Support
Agent
- Image Processing: Improved detection and extraction of data from images in PDFs
- Table Editing: Resolved problems with editing large Tables and targeting correct rows for changes
- Object Movement: Fixed AI-assisted Object movement
New Features
- Text within PDFs is now selectable
- Refinements to the new Index, including better Package collapsing by lane click
- Implemented “safe triangels” to right click menus in the Index
- Added Markdown support for Davinci Chat and Agent responses
Bug fixes
- Fixed bug related to adding References to Objects
- Fixed bug related to Search functionality
- Fixed image file size bug
- Fixed a bug preventing Tags from being opened from within Chat
- Stability improvements to both frontend and backend
Version 3.1.0 - Agent Toolboxes
New Features
Davinci Toolboxes
Toolboxes allow Davinci Agents to execute custom, user defined instructions- Davinci Agents can now use Toolboxes to assist with tasks
- Toolboxes allow Davinci Agents to execute custom, user defined instructions
- Users can create custom Toolboxes within Davinci
- Toolboxes are created using Python scripts
- Toolboxes can be exported as MCP Servers
- Users can test Toolbox functionality before allowing Agent use
- Agents will automatically choose appropriate Toolboxes when performing tasks
- Toolboxes can be deactivated to prevent automatic Agent use as desired
- Toolboxes and custom Code Objects can save files to the Project as Reference Objects
- Toolboxes and custom Code Objects can update existing Attribute Object values
Index Redesign
The Index has been reworked to add new functionality and a higher level of polish on existing functions- Rearranging Elements within the Index is now smoother and more consistent
- Redesigned the Drop Indicators when moving Index Elements
- Known Bug: Drop Indicators for Index Elements sometimes do not adhere to parent/child restrictions and will show invalid drop locations
- Redesigned the Drop Indicators when dragging external Files into the Index
- Files will now be uploaded to the nearest parent Package object
- File destinations are now indicated by a light box surrounding the Package that will contain the Reference Object after File upload
- File uploading now adheres to the parent/child restrictions imposed on Reference Objects
- Index Elements can now be dragged into the Workspace to open new Tabs
- Known Bug: If multiple Index Elements are dragged into the Workspace at once, only the first Object will be opened
Improvements
- Better overlay and feedback when dragging Files into Davinci
- Files can be dropped anywhere on the Workspace or Chat when in the Model Area to upload them directly to the Library
- Files can be dropped in the Index to upload them directly to a specific Package
- Inactive Files can now all be cleared at once with a new button in the Files Tab
- Note: Clearing inactive Files may cause compatibility issues with Snapshots that use them
- Read more about Inactive Files here
- Stability and Reliability improvements to Snapshots
- Known Bug: If a Snapshot takes more than 5 seconds to apply, the User may be redirected back to the Home screen until the Snapshot is fully applied. Re-open the Project once the Snapshot is fully applied.
- Improvements to Agent Display
- Agent Tool Inputs and Responses are now viewed directly in chat instead of a separate window
- Removed the split view previously used for Agent Responses
- Added a Checkpoint system to Agent Conversations
- Added ability for the User to Undo individual Agent Actions
Changes
- Removed ability to duplicate Reference Objects
Bug fixes
- Fixed a bug related to deleting multiple Model Objects at once
- Fixed a bug related to client-side Model Database traversal
- Fixed a bug related to resizing Workspace Panes via double clicking Tab headers
- Fixed a bug that allowed Reference Objects to be children of other Reference Objects
- Several fixes to internal command routing and AI performance
- Several backend fixes for performance and stability
Version 3.0.0 - Agents
Agent Capabilities
- Introduced Design Agent capabilities
- Added multi-agent support with concurrent agent execution
- Added agent tool calling
- Temporarily removed AI ability to create geometries and assemblies
Code & Output
- Improved AI generation of code
- Drastically improved code execution reliability
- Added ability to save figures from output as images
Document Features
- Improved AI document generation
- Overhauled manual editing in documents
- Added ability for AI to create documents with images
Table Features
- Substantial improvements in AI table creation and editing
- Tweaked manual table interactions
Presentation Features
- Added slide object
- Added ability for AI to create slides
File Management
- Added ability of agent to save files from a URL
- Added ability of agent to extract images from PDF files
- Added ability of agent to save figures from code output as images
Action System
- Rebuilt action definitions
- Added sequence view to actions
- Improved management of three.js instances
Project Area
- Added relationships to project area
- Tweaked project area views
- Tweaked project area file table
- Added upload file button to file tab
UI Improvements
- Enhanced collaboration features
- Improved right click menus