Recent Product Updates

We're excited to share the latest improvements to CreateAI! This release focuses on making it easier to get started, discover CreateAI experiences, and build AI project assistants faster.

Dashboard Improvements

We've refreshed the CreateAI dashboard to provide a cleaner, more intuitive experience.

What's new?

• Removed the large hero banner to give more space to the content that matters most.
• Added a personalized  experience for new and returning users.

• Introduced Quick Start, making it easier to discover: 

  Templates, 

  Learning Design Suite tools built with CreateAI (such as the Transcription Tool) and 

  Marketplace projects you can browse, borrow, and customize

• Moved Recent AI Projects higher on the page so you can quickly return to your work.
• Improved the dashboard layout and responsiveness, allowing Quick Start cards to display more cleanly across laptop screen sizes.
• Improved accessibility through updated color contrast and iconography.

 New FAQ Template

We've introduced a new FAQ Template to Quick Start. Simply upload your documentation or knowledge base to create an AI assistant that answers questions using your content.

The template includes sensible defaults to help you get started quickly while still allowing you to customize instructions, settings, and behavior at any time.

 Syllabot Improvements

The Syllabot template now uses GPT-5.5 by default, providing improved quality and performance for new Syllabot projects.

 Knowledge Base Improvements

We've continued improving the Knowledge Base experience with several usability enhancements, including:

• Improved table sorting behavior.

• Tooltip support for long or truncated file names.

   

• Backend indexing and synchronization improvements.

 API & Builder Improvements

Several enhancements have been made for developers and builders:

• API request forms now include clearer messaging when using Project Owner Tokens.

• Service Tokens are now the default selection when creating API requests.

   

• Hyperlinks within uploaded documents are now accessible to supported language models.

 User Experience Improvements

We've made numerous quality-of-life improvements across the platform, including:

• Faster project icon and banner updates during autosave.
• Improved Marketplace messaging when no results are found.
• Better project creation error messaging.
• Improved copy/share interactions.
• Updated model descriptions and experimental model messaging.

Bug Fixes & Performance Improvements

This release also includes numerous bug fixes, stability improvements, accessibility enhancements, and performance optimizations across CreateAI to improve the overall user experience.

May Release Notes - May 2026

New Features

• Auto Model Selection: CreateAI can now automatically recommend the best AI model for your project based on your instructions. This helps users get started faster without needing to compare and select models manually.

• Knowledge Base Redesign: 

  We've redesigned the Knowledge Base experience to make managing and using content easier. The update includes usability improvements and stronger Google Drive integration for a smoother knowledge management workflow.

• Claude 4.7 Opus Available: 

  Claude 4.7 Opus is now available in CreateAI, providing access to enhanced reasoning and advanced AI capabilities.

• AI Avatars (Pilot Availability): 

  CreateAI now supports AI Avatars for persona-based, multimodal interactions that combine conversation, voice, and avatar experiences. 

  This capability is currently available on a case-by-case basis for select pilots and use cases. If you are interested in exploring AI Avatars for simulations, learning experiences, or other interactive scenarios, please contact the CreateAI team.

• Expanded Integrations: 

  We've expanded support for Model Context Protocol (MCP) and enhanced CreateAI Builder APIs with OpenAI-compatible endpoints, making it easier to connect CreateAI projects with external tools, applications, and workflows.

• Deprecated model removed: Claude 4 Opus has been removed following its deprecation. This helps prevent errors and ensures only supported models are available.

Usability Enhancements:

• Project List Pagination: 

  Projects now load in pages, improving performance and navigation for users with a large number of projects.

• Improved Project Editing Experience: 

  The "Edit View" action has been relocated within the interface for improved discoverability and workflow consistency.

• Marketplace Search Improvements: 

  Marketplace search has been moved to the backend, improving search performance and scalability.

• Improved Chat Reliability: 

  We've made improvements to chat history, conversation loading, and overall chat reliability to create a more consistent experience.

• Analytics Dashboard Improvements: 

  The analytics dashboard has been updated to improve visibility and scrolling, making it easier to review project metrics.

• Smarter Agent Search Behavior: 

  Agent web search now uses a more efficient search strategy for simple queries, helping improve responsiveness.

• Universal System Prompt Support: 

  Added support for universal system prompts to provide more consistent behavior across projects.

Resolved Issues

• API Key Improvements: Fixed an issue where API keys could appear under the wrong project. and updated the API key request experience and user interface.

• Chat Experience Fixes: 

  Fixed an issue that could cause jumbled text when pressing Enter while responses were generating and f

  ixed an issue causing 404 errors when creating or opening new chats.

• Evaluation Tool Fixes: 

  Resolved backend issues affecting evaluation functionality and reliability.

• Project Management Fixes: 

  Fixed an issue that could cause errors when deleting older projects created before autosync was introduced. Also we added v

  arious improvements to project and memory management.

• Security Updates: 

  Implemented security enhancements for 

  Google Drive integrations, 

  Notifications and m

  emory-related functionality

• Additional Improvements: 

  Improved avatar responsiveness and listening behavior., Made backend updates to agent functionality and we have ongoing improvements to platform usability, reliability, analytics, performance, and security.

April Release Notes - April 2026 -

New Features

• API key access (in-platform)
  You can now request and manage API keys directly within CreateAI. Go to your project → Profile → API Keys → Request API Key and submit the form. Your API key will be available within 24 hours.
  Learn more: API Access Guide

• Cross-conversation memory
  CreateAI can now retain context across chats, leading to more consistent and personalized interactions.
  Learn more: Cross-Conversation Memory Guide

• AIR Platform (AI Research Acceleration Platform)
  CreateAI Builder now provides access to locally hosted models (e.g., Gemma 4, Llama 4 Scout) powered by ASU-managed infrastructure for secure and scalable AI use.

• New models available
  You can now use Gemma 4, Llama 4 Scout, Llama 4 Maverick, and GPT 5.5.

• Deprecated model removed
  Claude 3.7 Sonnet has been removed following its deprecation. This helps prevent errors and ensures only supported models are available.

• Banner image controls improved
  You can now reposition and resize banner images before saving, with a preview to ensure nothing important gets cut off.
  Recommended size: 1920 × 350 px

Usability Enhancements:

• Improved Model Selection: You can now scroll through models using your keyboard and press Enter to select.
• Improved Handling of Large Inputs: Text inputs larger than 30kb are now automatically converted into text files instead of returning an error.
• Auto-Save Improvements: Saving is now faster and smoother. You’ll see a “Saving…” status in the header, and the Publish button will be temporarily disabled while changes are being saved.
• Navigation & Interface Updates

1. Long project titles and subtitles now wrap correctly in banners.
2. “New Chat” replaces the previous “Reset” button label.
3. “Open as Viewer” option added to project actions menu.
4. “Created by” label removed from builder projects. 

• Project & Chat Improvements: Chats now load more reliably so older messages are visible, model switching is more consistent, and file uploads and date/time access are now enabled by default.

   

• Evaluation tool improvements: We’ve fixed issues where questions appeared across projects, reappeared after deletion, and where evaluations could get stuck loading.

• Accessibility Improvements: 

  We’ve improved accessibility across CreateAI to make the platform easier to use for everyone, including better color contrast, clearer form labels, and improved keyboard navigation.

Resolved Issues

• Fixed LaTeX and markdown rendering issues in queries and responses
• Fixed issue where pressing Enter cleared the project icon
• Updated duplicate project name messaging (removed non-functional “replace” option)
• General fixes to improve reliability and performance

March Release Notes - March 2026

New Features

• New Models Available
  You can now use Gemini 3.1 Pro, GPT 5.3 Instant, GPT 5.4 (Experimental), GPT 5.4 Mini & Nano, and Nano Banana Pro 2 (API).
• Marketplace Improvements
  Project owners now have more control and visibility when sharing your projects in the Marketplace. You can choose whether others can view your instructions and make a copy of your project, add a category to help others discover it more easily, and view more detailed information about projects directly within the Marketplace.
• Domain-Specific Accuracy (Evaluation)
  Project owners can now evaluate how accurate your AI project is within its intended subject area. By adding sample questions and expected answers, CreateAI will assess responses and provide a score to help you understand how well your project performs. Use these insights to improve your knowledge base and refine responses.
• Project Personalization
  You can now customize your project with a banner image and icon using your own images or the suggested library.

• Media Library Update
  You can now select multiple files at once to delete them more easily.

   

• File Support Update
  You can now upload and use .ps1 files in CreateAI Builder and Chat.

Usability Enhancements:

• Auto-Save Updates
  Your project now saves automatically as you make changes, so you don’t lose work.
• Default Model Update
  New projects now start with GPT 4.1 Mini by default.
• Canvas Improvements
  Content in Canvas now refreshes automatically to stay up to date.
• Improved Navigation and Access
  You’ll now see clearer messages when accessing restricted pages.
• Model Selection Improvements
  You can now scroll through models using your keyboard and press Enter to select.
• Improved Handling of Large Inputs
  Text inputs larger than 30kb get converted into text files before submitting instead of the previous experience, where you would eventually see an error due to the content being too long.
• Cleaner Experimental Model Experience
  Messaging around experimental models has been simplified for clarity.

Resolved Issues

• Canvas Embedding Fix
  Projects can now be embedded in Canvas without issues.
• User Management Fix
  Adding or removing users from a project now updates immediately without needing a refresh.
• Auto-Save Typing Fix
  Text fields no longer lose focus while typing.
• Slider Behavior Fix
  Model setting sliders now snap correctly to valid values.
• General Stability Improvements
  We’ve made several improvements to make CreateAI more reliable and smooth to use.

Product Release Notes - February 2026

New Features

• Gemini 3.1 Pro Now Available: Gemini 3.1 Pro has been added as a new model option in the platform.

• Improved Loading Feedback: You’ll now see a “Still thinking…” message if a response takes longer than expected, giving clearer feedback during generation.

• Video File Support: You can now upload MP4 and WEBM files in the CreateAI Chat Media Library.

• Direct Privacy Access: The “Privacy settings” link in your profile now opens cookie management directly.

• URL Auto-Sync Controls: You can now choose how often URL sources are automatically refreshed (Daily, Weekly, Monthly, or None).

Usability Enhancements:

• Duplicate Upload Prevention: You can no longer upload the same file multiple times accidentally in the in-context uploader.

• More Specific Error Messages: Failed file uploads in the Knowledge Base now display clearer, more helpful error messages.

• Accurate “Last Published” Timestamp: The timestamp now correctly updates when files are removed from the Knowledge Base.

• Improved Print Chat Experience: Saving chats as a PDF no longer cuts off content after the first page.

• Model Display Improvements: Model names now display correctly after refreshing the chat page.

• Mobile Model Selector Fix: The model dropdown no longer appears underneath the keyboard on mobile devices.

• Feedback Form Improvements: 

  Uploading photos no longer requires a double click and the 5-photo upload limit is now properly enforced.

• Markdown Preservation: Markdown formatting is now preserved across Builder and Chat responses.

• Improved Layout Spacing: Spacing and alignment have been improved on larger screens.
   

Accessibility Improvements

We’ve made several updates to improve keyboard navigation and screen reader support:

• Improved nested interactive controls in CreateAI Chat and Media Library.
• Improved color contrast in dropdowns and Agents features.
• Updated ARIA attributes for the Agents toggle.
• Improved keyboard access for scrollable regions.
• Enhanced screen reader labeling across key components.

Resolved Issues

• Fixed an issue where pressing Enter could interrupt response generation.
• Resolved an issue where Syllabot messaging persisted when switching between projects.
• Fixed duplicate drag-and-drop upload behavior.
• Improved URL ingestion fallback reliability.
• Improved upload timeout handling for stalled files.

Product Release Notes - January 2026

New Features

• Knowledge Base Auto-Save Indicator: A new “Auto-saved” message now appears in the top-right corner of the Knowledge Base page for added clarity.

• Expanded File Support: You can now upload and use the following additional file types in the Knowledge Base, Google Drive, and in-context uploads: .vtt (Web Video Text Tracks) .stt, .v (Verilog), .sv (SystemVerilog), .yaml, .toml, .tex (LaTeX), .tf (TerraForm). 

• Copy Enhancements: You can now copy the prompt, response, or both in one click with new dedicated buttons.

• Model Expansion: Gemini 3 Flash and Pro are now available in the platform.

• Dashboard Flexibility: Dashboard items now support standard open-in-new-tab behavior.

• WEBP File Uploads: Users can now upload .webp image files into both CreateAI Builder and CreateAI Chat

Usability Enhancements:

• Upload Without Prompt: You can now upload a file without entering a prompt. The system will handle it by default with the instruction: “User has uploaded a file. Please follow the instructions in the system prompt.”
• Improved Button Feedback: The Regenerate and New Chat buttons are now greyed out while a response is being generated to prevent unresponsive clicks.
• Syllabot Template Improvements: The “Go to knowledge” button and related messaging now persist until at least one file has been successfully uploaded.
• Auto-Publish Projects: Toggling ASU-wide access now automatically publishes the project.
• Faster Marketplace Browsing: Marketplace now shows 50 projects per page by default instead of 10.
• Model Search Is Back: You can once again search by name in the model selection dropdown.
• Markdown-Friendly Copying: Markdown formatting is now preserved when copying responses from CreateAI Builder into tools like Google Docs.
   

Resolved Issues

• Persistent Tooltips: Fixed lingering tooltips on CreateAI Chat and CreateAI Builder viewer message buttons.
• Duplicate File Uploads: Clearer error messaging when attempting to upload the same file to local files or Google Drive.
• Mobile Experience: Fixed an issue where creating a new project on mobile could show a blank screen.
• Custom Instructions: Fixed an issue where copy/paste or select all/backspace wouldn’t trigger the Save button.
• “Report a Problem”: Fixed an issue where clicking this on the viewer side reset the current chat.
• Share Tab Scroll: Fixed the horizontal scrollbar issue in the “Users with access” section on smaller screens.
• Stuck Uploads: Files stuck in "Uploading" state for over 15 minutes will now fail, allowing the user to delete them.
• URL Upload Reliability: ASU URLs now correctly fall back to Firecrawl if iSearch is unavailable.
• Mobile UI Fix: Resolved an issue where the model selector overlapped the Save button on mobile.
• Accepted File Types Cleanup: Removed unsupported .MOV files from KB accepted types.

Product Release Notes - December 2025

New Features

• Canvas Integration for CreateAI Builder
  Faculty can now add AI projects directly to Canvas courses. More details on this feature Here 

   

• New "Report a Problem" Button: Users can now quickly flag technical issues directly from the viewer interface, thereby making it easier to get support when needed.
• New Models: Users can now choose from the latest models which include Gemini 2.5 (Flash, Flash Lite, Pro), Claude 4.5 (Haiku, Opus, Sonnet), and GPT (5.1 & 5.2).
   

Usability Enhancements:

• Easier Save & Publish Flow
  Users can now clearly see the status of their app: 

  Greyed-out Save = all changes saved, Yellow Save = unsaved changes and 

  Green checkmark = ready to publish

• Better AI Responses by Default: Users can now start new AI projects with smart default settings (Top K = 15), ensuring quality responses out of the box.
• Helpful Mobile Tooltips: Users on mobile now see clear guidance when a file can’t be processed, making it easier to resolve upload issues.
• Smarter Syllabot Setup: Users are now reminded to upload a syllabus before configuring their AI. The reminder goes away automatically once a document is uploaded.
• Analytics Are In: Users’ interactions across the Compare Tool, CreateAI Chat, and App Viewer are now tracked (via Google Analytics and Tag Manager), which will help our  team improve performance and usability.
• Improved Voice Input: Users can now pause and resume their mic without losing previous input, and easily delete or re-record spoken text for better voice-to-text control.
• Better Readability: Users now benefit from improved color contrast throughout CreateAI Builder, making text easier to read and enhancing accessibility.
   

Resolved Issues

• CSV Uploads Now Work Reliably: Users can now upload all CSV files without failure or freezing due to previous decoding issues.
• Cleaner Knowledge Base Columns: Users can now view consistent column names and order across the Files, URLs, and Google Drive tabs for easier scanning and sorting.
• Mobile Tooltip Glitch Fixed: Users on mobile will no longer see tooltips that stay stuck after scrolling they now disappear as expected.
• Chat Tooltip Cleanup: Users will now experience smoother chat interactions tooltips disappear properly when scrolling or switching tabs.

Accessibility Updates

• Fixes to Pop-Ups (Modals): Users can now navigate the "Report a Problem" and "Give Feedback" dialogs more easily with screen readers and keyboards.
• Improved Sliders: Users relying on screen readers can now hear labels and values for the Temperature and Top K sliders in the Behaviors tab.
• Keyboard-Friendly Compare Tool: Users can now scroll through Compare results using standard keyboard controls, improving navigation.
• Cleaner Tab Navigation: Users can now more easily navigate tabs using screen readers and keyboards, thanks to improved structure and Web Accessibility standards compliance.

Product Release Notes - November 2025

New Features

• New in-app notification banners: Users will now see time-stamped alerts across key pages (e.g., Dashboard, Builder, Marketplace) to stay informed about important updates, issues, and resolutions.
• Expanded URL ingestion: You can now ingest any URL into your project’s Knowledge Base. ASU URLs will use iSearch by default, with Firecrawl as a fallback. All other URLs use Firecrawl directly.
• Project Analytics Dashboard: Users can now access project-level analytics, including engagement and project activity metrics, all in one place in CreateAI Builder. Note: API-generated projects are not yet included; support is being explored.
   

Usability Enhancements:

• Improved feedback experience: Users now see a confirmation message after reporting a message in CreateAI Chat.
• Added file type help link: Users can now click a “See all accepted file types” link in the CreateAI Builder to view supported formats in a help article.
• Improved prompt validation: The "Generate" button in the CreateAI Compare Tool is now disabled until a prompt is entered, preventing empty submissions.
• New tooltip for partially processed files: Users will now see a tooltip indicating if a file wasn't fully uploaded, with guidance to re-upload as a PDF or in smaller parts.
   

Resolved Issues

• Fixed mobile link styling: Resolved an issue where links appeared bright blue on mobile devices in CreateAI Chat and CAB viewer; they now match expected styles.
• Fixed incorrect project status display: Projects are no longer marked as “Live” when adding viewers unless explicitly published, preventing access errors and confusion.
• Fixed project tab navigation: Projects now consistently open to the Basic Settings tab, even when revisited in the same browser tab.
• ​​Resolved Firecrawl issue: Fixed a backend bug that prevented depth 0 (root-level) URLs from being crawled correctly using Firecrawl.
• Fixed responsive layout issue: The Knowledge Base header now displays correctly on smaller screens without being cut off.
• Fixed mobile chat renaming issue: Users can now successfully edit chat names on mobile without the sidebar collapsing or losing keyboard focus.
• Improved input validation for sharing: Users can now only share projects with valid ASURITE usernames or emails, preventing errors caused by null or incorrect entries.
• Resolved Google Drive file refresh error: Fixed an issue where files with slashes ("/") in their names caused refresh failures in the Knowledge Base.
• Fixed mobile notification banner layout: Resolved overlapping issues between the sidebar icon and notification banner, improved text visibility, and ensured banners stay pinned on scroll.

Product Release Notes - October 2025

New Features

• Marketplace Launch: Users can now share their projects with the ASU community via the new CreateAI Builder Marketplace. Projects marked “Viewable by everyone at Arizona State University” will appear automatically. Users can manually hide projects as needed.
• Claude 4.5 Haiku Model Integration: The Claude 4.5 Haiku large language model is now available in CreateAI, offering improved performance and additional model choices.
• Image Paste Support: Users can paste images directly into text fields across CreateAI Chat and CreateAI Platform apps with  no upload needed.
• Role-Based Filtering for Projects: A new filter has been added to the “Role” column on the MyAI Projects page, allowing users to quickly sort projects by roles such as Owner or Editor. This enhancement helps users manage and locate projects more efficiently.
   

Usability Enhancements:

• Project Sharing UI Improvements: Users can no longer add duplicate ASURITEs, and role drop-downs remain accessible even with many users. The delete function now works consistently.
• Scrollable Sidebar Navigation: A scrollable sidebar now prevents overlapping menu items, improving readability and accessibility. 
• Improved Long Project Name Display: Long project names now wrap or truncate correctly in both the sidebar and the Analytics Dashboard, maintaining a clean and readable layout.
• Mobile Interaction Enhancements: Tabs now respond on the first tap, eliminating the need for double-tapping, and the “New Chat” icon is fully visible and accessible across all devices.
• Copy Button Stability: The interface no longer shakes on hover, and tooltips now display correctly. 
   

Resolved Issues

• OpenAI Retry Logic: New retry system now handles API rate limits, timeouts, and intermittent issues, improving stability.
• Media API Upload Behavior: File upload status now displays “updating” until processing is complete, ensuring accurate status feedback.
• Improved File Upload Reliability: Users can now successfully upload Excel documents to the Knowledge Base without errors, thus ensuring a smoother and a more consistent data integration.

Product Release Notes - September 2025

New Features

• Google Drive Sync for Knowledge Base: You can now sync your Google Drive files with the Knowledge Base to keep documents up to date. Sync changes manually for individual or all files, and see clear error messages for issues like deleted or renamed files. This works with your existing Google connection and no setup is  needed.
• Smarter Source Awareness in KB
  The AI now knows which file each answer came from and can cite the filename or URL when asked. This works automatically for new uploads. (To apply it to older files, re-upload them.)
• New Production Link Available: We officially moved from beta to production The platform is now live at: platform.aiml.asu.edu . This  accessible to all ASU faculty and staff.

Usability Enhancements:

• Improved Chat Stability: To ensure smooth performance in long chats, older messages are now automatically condensed, while recent messages remain fully intact. This helps maintain context and improves overall conversation stability without losing your important history.
• Improved model selection experience by removing deprecated options to reduce confusion and improve performance.
• Larger Custom Instructions Box: The custom instructions box now expands to a larger, more readable size when clicked. This makes it easier to write and review longer instructions. The expanded view works smoothly across screen sizes and retains your input when toggling between views.

Resolved Issues

• Deprecated Models Removed: As of Anthropic’s deprecation of Claude 2, Claude 2.1, and Claude Instant: These models have been removed from the model selector in CreateAI Builder, CreateAI Chat, and CreateAI Compare. This resolves the issue where users experienced errors when selecting now-unavailable models.
• File Uploads Now Support Special Characters:
  You can now upload files with special characters (like +, @, #) in the name without errors. No UI changes were made.
• Fixed: Users were unable to submit the Report a Problem form due to a false error saying “Please select a project” even when a project was selected. This issue is now resolved. You can submit the form once a project is selected.
• Fixed Chat Feedback Form Submission Fixed: Users can now successfully submit chat feedback reports. A bug previously blocked submission with a “Rating is required” error, even when a rating was selected. This has been fixed as ratings are now recognized correctly, and comments remain optional.
• The full screen button in CreateAI Compare now works as expected. Clicking it will properly expand the selected model’s response for easier viewing. This fix improves both desktop and mobile behavior, though mobile layout improvements are planned for a future update.
• Fixed: File Upload Error When Switching Chats: Previously, uploading a file in one chat and then resetting or switching to another caused an error. This is now fixed as uploaded files and prompts are properly cleared when starting a new chat or navigating between chats, preventing submission issues.
• Fixed: Logout Now Fully Logs Users Out: The Logout button on CAB now works as expected. Previously, users saw a "signed out" screen but remained logged in due to cookies not clearing. This fix ensures users are fully logged out and prompted to sign in again on both desktop and mobile.
• Fixed: Model Names Now Stick to Original Messages: Previously, when switching models mid-chat, older messages would incorrectly update to show the new model name. This is now fixed as  each response now retains the model name used at the time it was generated, even if you change models later. This applies to both default model names and custom response names.
• Fixed: Top-K Slider Sends Valid Minimum Value: Resolved an issue where the Top-K slider sent 0 to the backend when set to the far left. It now correctly sends a minimum valid value (e.g., 1), keeping UI and backend in sync.