Sprint 33: March 24- March 30, 2025
Changelog: March 31, 2025
This update introduces a new visual indicator for debugging and several quality-of-life improvements for navigation and configuration. It also fixed many bugs to enhance stability, billing accuracy, and user interface consistency.
✨ New Features
Add Debug Icon Blinking Animation
The debug icon will blink whenever a debugging session is in progress to provide a clearer, more immediate visual cue. This helps you instantly identify when debugging is running without needing to check other status indicators.
🚀 Improvements
We’ve made the OAuth authentication process more intuitive and error-proof. The blue “Authenticate” button in the builder will now only appear when the required token and secret fields are properly populated. The grey sidebar button remains visible but will provide clear error messages if clicked without the necessary configuration. This change prevents failed authentication attempts and provides better guidance.
We’ve implemented a “Collapse All” button in the components tab to improve usability and consistency. This update brings the same convenient functionality from the integration tab, making it easier to navigate and manage complex component hierarchies.
Implement Consistent Navigation Elements in Enterprise Plan Pages
The sidebar and top navigation bar have been restored across all Enterprise Plan pages. This change provides a more consistent and familiar user experience, allowing for seamless navigation between different sections of the platform.
🐛 Bug Fixes
Fixed an issue where the debug tooltip would overlap and obscure component settings. This allows you to properly view and modify settings during debugging, improving your development and configuration workflows.
Component Name Alignment
Corrected an alignment and truncation bug with component names, which enhances user interface consistency and readability.
Enterprise Template Crown Icon
Addressed a bug causing the crown icon for Enterprise templates to be incorrectly sized and positioned. This improves the user experience by making Enterprise templates easier to identify at a glance.
Incorrect Page Title on Templates Page
Resolved a bug where the Templates page was incorrectly displaying the “Home” page title, which eliminates user confusion when navigating the application.
Fixed a critical billing issue where extra seats were not being included in monthly Stripe invoices. This prevents revenue loss by ensuring customers are charged correctly for all additional seats.
User Management Role Assignment
Resolved a bug in User Management where editing a team member’s role would incorrectly assign the default role. Administrators can now properly assign specific roles and manage permissions as intended.
Removed Deprecated Plan Pages
Outdated pricing plan pages have been removed from the production environment. This is critical for revenue protection as it prevents users from subscribing to deprecated plans.
Serverless Code Component in Analytics
The Serverless Code component usage statistics are now correctly grouped within the main “Code” section on the Analytics page. This organizational fix makes it easier to monitor and compare related components efficiently.
Inconsistent Intro Message
Fixed an issue where the intro message set in agent settings was not being reflected in the chatbot’s behavior. This resolves user confusion and ensures the chatbot’s functionality is predictable and aligns with your configuration.
Sprint 34: March 31- April 6, 2025
Changelog: April 7, 2025
✨New features
No new features in this sprint.
🚀Improvements
No improvements in this sprint.
🐛Bug fixes
No bug fixes in this sprint.
Sprint 35: April 7- April 13, 2025
Changelog: April 14, 2025
In this sprint, you’ll enjoy new controls and testing capabilities, alongside a host of improvements focused on UI consistency and user guidance. We have also addressed a significant number of bugs to enhance stability, improved the debugging workflow, and fixed visual regressions.
✨ New Features
Weaver Daily Limit
We have implemented a daily usage limit for the Weaver feature. This will help control usage patterns and manage resource allocation, preventing system abuse and ensuring fair access.
Add Event to Set Mock Values
A new event functionality has been added that allows you to set mock values via a specified chat link. This will enable easier and more efficient testing and simulation by providing a direct mechanism to set mock data through events.
🚀 Improvements
Debug Inject UI Improvements
We’ve rolled out the first version of improvements to the Debug inject user interface, enhancing the usability and functionality of the debugging tools for developers.
All sidebar titles now follow a consistent format and style, and each sidebar includes an appropriate icon. This improves visual consistency and makes navigation more intuitive.
Clearer Language in Advanced Fields
The wording in the advanced fields has been changed from “description” to “behavior,” with updated headers and placeholder text to align with agent settings. This provides clearer guidance when configuring component behaviors.
Improved UI for Debugging Sessions
We’ve enhanced the builder’s UI during debugging sessions with clearer visual states for components and better process indicators, creating a less confusing and more efficient workflow.
Consistent Page Title Positioning
Page titles across the Data Pool, Account, User Management, and Manage Roles pages have been realigned to match other sections, creating a more uniform look and feel across the application.
Mock Values in API Key Modals
Example/mock values have been added to the API key configuration modal. This provides clear guidance on the expected input format, making setup easier for users.
Clearer Upgrade Path for Custom Plans
The upgrade experience for users on early adopter custom plans has been clarified. Vague notifications have been replaced to ensure users have clear instructions on how to proceed with upgrading their subscriptions.
Feedback for Non-Owner Subscription Cancellation
The system now provides clear visual feedback and an error message when a user who is not the team owner attempts to cancel a subscription. Now, users will no longer be confused about permission restrictions.
Persistent Error Messages for Integrations
Red error boxes for integration components with missing parameters will now remain visible instead of flashing and disappearing, ensuring users can see and address configuration issues.
🐛 Bug Fixes
Serverless Code Component Icon Display Issue
We fixed an issue where the icon for the serverless code component was not displaying correctly. This ensures the correct visual representation for these components, resolving potential confusion and making them easier for you to identify in the UI.
Alignment Issue with “(thinking)” Label in Weaver Text Generation
We corrected a CSS alignment problem with the “(thinking)” label in the Weaver component. The update fixes visual inconsistencies, ensuring you experience a polished and properly aligned interface during text generation.
Debugger Attach Button Remains Disabled After Connection Failure
Previously, when a debugger connection failed, the “Attach” button would remain disabled. We’ve fixed this so you can now make subsequent connection attempts without having to refresh the page, preventing disruptions to your workflow.
Border Overlapping Component Name on Selection
We fixed an issue where the selection border would overlap and partially cover a component’s name. This improves the readability and visibility of component names when they are selected.
Agent Security Settings Save Failure in Non-Organization Spaces
We resolved a bug where security settings for agents would fail to save in any space other than the default organization space. You can now properly configure security settings in all your spaces, removing this limitation.
Logs Detail Page – Log Loading and Download Features Failing
We fixed two critical issues on the Logs Detail page where both the ‘Load full content’ and ‘Download log’ features would fail for large logs. You can now access and download complete log information, restoring your ability to troubleshoot issues and analyze agent performance.
Chatbot Component Failing to Load in Specific Agents
We resolved an issue that caused the chatbot component to fail to load in certain agents. This restores core functionality, ensuring you can interact with the chatbot in all affected agents.
Input fields in the form preview were not resizing properly to accommodate long text. We’ve fixed this to improve usability, ensuring your text is not cut off and is easy to edit.
Fixed a bug where the tooltip overlay in the debug inject feature would cover component settings. You can now properly access and interact with these settings when the tooltip is active, improving your ability to configure components.
The chat interface in Agent Weaver was not automatically scrolling down to show the action buttons at the end of the creation process. This has been fixed to ensure you won’t miss important next steps, as these critical buttons are no longer hidden from view.
Builder Selects Component Borders Missing Regression.
Fixed a visual regression where select components in the Builder were missing their borders. This makes them easier for you to identify and interact with, thanks to the restored visual boundaries.
The close button on the “New Features” modal was positioned outside the viewport on mobile devices, making it impossible to close. This has been fixed, removing a significant usability barrier and allowing mobile users to continue using the application.
False “Missing settings” Error Message in Integration Components
Integration components were incorrectly displaying “Missing settings” error notifications even when they were fully configured. This fix eliminates that confusion, ensuring you don’t waste time troubleshooting configurations that are already correct.
Component Settings – Red API Key Error Notification Persists After Key is Saved
After saving an API key, a red error notification would incorrectly reappear when opening the component settings sidebar. We’ve fixed this to create a more consistent and less confusing experience, ensuring the UI correctly reflects that the API key has been saved.
Empty Call Tree Modal After Downloading Logs
The call tree modal would appear empty if you clicked ‘Download Logs’ before ‘Show Call Tree’. We’ve fixed this so you can now view the call tree information regardless of your workflow, ensuring you can access important debugging data when you need it.
Sprint 36: April 14 – April 20, 2025
Changelog: April 21, 2025
This sprint delivers powerful new workflow efficiencies, including keyboard shortcuts and direct canvas editing for agents. We’ve also shipped several UI improvements for better clarity and responsiveness and resolved a wide range of bugs affecting subscriptions, integrations, and the debugging experience.
✨ New Features
Keyboard Shortcuts for Layer Ordering
You can now rearrange the layering of Notes and other components more efficiently with keyboard shortcuts. Use Ctrl/Cmd + Shift + ] to bring elements to the front and Ctrl/Cmd + Shift + [ to send them to the back, all without using menus.
Resizable Note Component
We’ve enhanced the flexibility of the note component. You can now resize notes not just horizontally, but also by dragging the bottom-right corner, giving you more control over your canvas and demo layouts.
Agent Card in Canvas (MVP)
A basic agent card has been implemented on the canvas. Clicking it opens the agent’s settings directly in the right sidebar, allowing you to access and modify configurations without leaving your current workflow.
🚀 Improvements
‘Chat with Agent’ UI Polish
Addressed multiple UI issues in the chat interface, including inconsistent borders and uneven margins, to create a higher quality and more consistent user experience.
The generic ‘Help’ tooltips for the Code and Serverless Code components have been replaced with more descriptive text. This helps you better understand the differences and use cases for each option.
The right sidebar has been updated for better responsiveness. “smythos model rates” has been changed to a more intuitive “$ view pricing” button, and a new tooltip explains the pricing model more clearly.
The OAuth “Authenticate” button will now only appear when all required configuration fields are populated. Clicking it with an incomplete setup will trigger a clear error message, preventing failed attempts.
🐛 Bug Fixes
Fix role dropdown in user management
We fixed an issue in the user management dropdown menu where the placeholder text ‘Choose a role..’ could be selected as a valid option. This prevents users from selecting an invalid role, improving data integrity.
Subscription System Blocking Purchases After Failed Payment
We resolved a critical bug where the subscription system would block users from making a new purchase after a payment failed. This removes a frustrating barrier, ensuring users who experience an initial payment failure can successfully complete their subscription purchase.
SmythOS Plan & Plan Price Not Syncing to Hubspot
We addressed a sync issue where SmythOS Plan & Plan Price properties were not updating correctly in Hubspot. This fix ensures customer plan information is accurately tracked, eliminating the need for manual updates.
Agent Settings Authentication Method Display Inconsistency
We fixed a bug where the UI dropdown in Agent Settings failed to display the correct authentication method after saving. This eliminates user confusion and ensures you can clearly see which authentication method is active.
404 Error on Staging Documentation Pages
Documentation pages in the staging environment were consistently returning 404 errors. We’ve resolved this, restoring access to the content and allowing for proper testing and verification before release.
OAuth Configuration Displays Incorrect URLs for Custom Subdomains
We fixed a major issue where the system generated incorrect authentication URLs for agents deployed with custom subdomains. This allows users with custom subdomains to successfully configure OAuth authentication.
The reload button in the Test sidebar was incorrectly visible when switching to the Form tab. We’ve removed it from contexts where it doesn’t apply, creating a cleaner and less confusing UI.
Agent Endpoint Logs Not Captured in Debug Bar
We resolved an issue where agent endpoint logs were not being captured in the debug bar. This restores important diagnostic and monitoring capabilities, making it easier to track and analyze operations.
Agent Weaver – Incorrect error message when free credits exhausted
The system was displaying a misleading ‘daily limit’ error message when users exhausted their free credits. We’ve corrected this to ensure you receive accurate information about your account limitations and credit system.
Empty Advanced Section Displayed for FLUX Models in Image Generator
We fixed a bug where an empty “Advanced” section was displayed for FLUX models in the image generator. This removes UI clutter and reduces user confusion.
OpenAPI Integration – Component Not Loading After Providing Manifest URL
We resolved a problem where the OpenAPI component would fail to load after a manifest URL was provided. This fix enables you to successfully integrate OpenAPI components into your workflows.
Sprint 37 April 21 – April 27, 2025
Changelog: April 28, 2025
Expect significant new capabilities in this new sprint, including plan-based usage limits for Agent Weaver, PDF support in chat, and Stripe abandoned cart metrics. We have also shipped major improvements to the pricing plan UX and unified the serverless code component, alongside crucial bug fixes to improve platform stability, collaboration, and the user interface.
✨ New Features
Usage Limits for Agent Weaver Based on Account Type
We have implemented a credit-based usage limit system for Agent Weaver. This new model provides free users with 5 requests per day and paid users with unlimited requests, creating a sustainable usage model and a clear upgrade path for users who need additional capacity.
PDF Support in Chat
You can now handle PDF documents directly within chat interactions. This feature allows users to upload, reference, and discuss PDF documents within conversations, significantly enhancing document-based collaboration.
Stripe Abandoned Cart Metrics
For our Stripe integration, we have implemented metrics to track abandoned shopping carts. This provides visibility into customer behavior during the checkout flow, allowing for better analysis of potential revenue loss.
🚀 Improvements
Pricing Plan UX Improvements
We’ve enhanced the user experience of the pricing plan interface, improving usability and clarity to help you make more informed decisions and potentially increase conversion rates.
Unified Serverless Code Component
Instead of maintaining separate components for different plans, we have created a single, unified serverless code component. This component uses plan-based access controls, which simplifies the codebase, provides a clearer upgrade path, and enables more targeted usage reporting.
API Endpoint Skills UI/UX Improvements
The workflow for API Endpoint Skills has been streamlined. We’ve added a ‘Test API Endpoint’ button in the Skill sidebar, enabled plaintext input for endpoint names, and improved how the sidebar closes when errors are present for a more intuitive experience.
🐛 Bug Fixes
We’ve corrected tooltip misalignment in the builder page sidebar. This fix ensures tooltips are properly positioned next to their corresponding icons, which reduces user confusion when navigating the interface.
Service Incorrectly Paused After Free Credits Depletion
We fixed an issue where the service was incorrectly paused after free credits were depleted, instead of transitioning to billable credits. This prevents unexpected service interruptions and ensures a seamless experience for users.
Spaces Bugs
We have addressed several unspecified bugs related to the Spaces feature to improve its overall stability and performance.
Weaver Next Steps Showing After Every User Message
The Weaver “next steps” interface was incorrectly appearing after every user message. We’ve fixed this to create a less disruptive and more intuitive experience for you.
We fixed a bug where the API Call component would fail to process binary input data. This restores the ability to make API calls with binary data, enabling file uploads and other critical data transfer operations.
Member Invitation Errors for New Accounts
We resolved multiple ‘Setting not found’ errors that were blocking collaboration features for new accounts. This important fix ensures new users can now fully utilize all collaboration tools, including space invitations and agent sharing.
First Name Property Incorrectly Auto-Updated to ‘User’ by Product App
We addressed an issue where the product app would incorrectly auto-update a contact’s first name to ‘User’ in HubSpot. Your manual changes will now persist correctly, ensuring data integrity.
Sprint 38: April 28 – May 4, 2025
Changelog: May 3, 2025
This sprint significantly expands our integration capabilities with the introduction of Amazon Alexa support and the Model Context Protocol (MCP). We’ve also added several new state-of-the-art LLM models. Key improvements include a major security enhancement to our vault storage and numerous UI refinements, alongside critical bug fixes that address core functionality, platform stability, and user experience.
✨ New Features
Alexa Integration for SmythOS Agents
We have implemented an integration with Amazon Alexa, allowing you to interact with your SmythOS agents via voice commands on any Alexa-enabled device. This expands accessibility into voice-controlled environments and enables new use cases for home automation and hands-free scenarios.
MCP Integration for Standardized Communication
SmythOS now supports the Model Context Protocol (MCP). This two-phase implementation includes a new MCP component and an MCP embodiment for agent exports, enhancing interoperability by allowing agents to connect with third-party platforms that support the protocol.
Support for New LLM Models
We have integrated five new cutting-edge language models: Gemini 2.5 Pro, GPT-4.1, GPT-4.1 mini/nano, and Sonar Deep Research. These models are available in both the LLM Assistant and GenAI LLM components, providing you with more powerful and flexible options to customize agent performance.
Book Intro Call in Onboarding Flow
For users on our Startup, Scaleup, and Enterprise plans, we’ve added a step in the onboarding flow to schedule an introductory call. This provides higher-tier customers with guided onboarding to improve activation and success.
🚀 Improvements
Vault Storage Modernization to Exclusive HashiCorp Implementation
We have migrated our vault key storage from a dual-storage approach to a single, HashiCorp-only implementation. This enhances security by reducing attack surfaces and eliminates data redundancy, all with no visible changes to the user interface.
Relocated “Computer Use” Component
To improve interface organization, the “Computer Use” component has been moved to the “Tools” section, placing it in a location that better aligns with its purpose and usage patterns.
Workflow Completion Status Message
The application now displays a clear confirmation message indicating whether a workflow has finished successfully or encountered an error, providing better feedback on execution status.
UI Refinements
- Vault Modal UI: Minor tweaks have been made to vault modals to improve visual consistency across the application.
- Builder UI Design pt 2: We’ve continued making refinements to the Builder UI to enhance the overall user experience.
🐛 Bug Fixes
We fixed an issue in the staging environment where the documentation sidebar was showing raw HTML code instead of properly formatted navigation links. This makes the documentation system usable again, allowing you to navigate content and improving the user experience.
Multiple Embodiments Integration – Critical Destructuring Error
We resolved a critical destructuring error related to the ‘planName’ property that caused agent skills to fail across all embodiments (Alexa, Agent LLM, MCP). This fix restores core functionality, allowing you to execute agent skills successfully across all integration points.
Toast Notifications Not Displaying in Staging Environment
We fixed an issue where toast notifications were failing to display in the staging environment. You will now receive proper confirmation and feedback for your actions, improving usability and the overall user experience.
Agent Weaver Service Interruption After Free Credits Depletion
We resolved a service interruption for paid users where Agent Weaver would stop working after free credits were depleted instead of transitioning to regular billing. This fix prevents you from being blocked from your work and ensures seamless continued service.
Access control bugs
We addressed several issues with the access control system to resolve potential security vulnerabilities and ensure user permissions and system access function as intended.
Agent Auth Settings Save Failure
We fixed a bug where OAuth settings could not be saved from the agent-settings page, which also failed to provide error feedback. You can now successfully configure agent authentication settings and will receive clear status updates.
API Component – Array Formatting Issue in Body Payload
We fixed an issue where square brackets were being removed from JSON arrays in API Call component body payloads. This prevents malformed requests and ensures your API integrations that use arrays will no longer fail due to invalid formatting.
Agent Card State Management Issues in Builder
We resolved several inconsistent state management issues in the Builder that affected real-time synchronization. This fix ensures that changes to agent settings and deployment status are correctly reflected across all UI components and tabs.
We fixed failures with the ‘Fix with AI’ button in the staging environment. This restores the ability to automatically resolve component errors for both integration and code components, improving the effectiveness of your error resolution workflow.
Sprint 39: May 5 – May 11, 2025
Changelog: May 12, 2025
In this sprint, the focus is on enhancing platform stability and improving the user experience. While there are no new features, we’ve implemented significant improvements, including adjusted trackpad sensitivity in the Builder and a comprehensive regression test suite for our GenAI models. We have also addressed several critical bugs related to billing, integrations, and the credit system.
🚀 Improvements
Adjust Trackpad Zoom Sensitivity in Builder
We’ve adjusted the zoom sensitivity for trackpad users in the Builder. The previous ratio was too high, causing exaggerated zoom effects. This change makes navigation more precise and aligned with industry standards like Figma.
SRE Regression Test Suite for GenAI Models
A comprehensive regression test suite has been implemented for our GenAI LLM models (including custom models, GPT, Claude, and Gemini). This suite verifies model functionality with and without image inputs, checks usage reporting, and confirms credit exhaustion error handling, ensuring system reliability and a stable user experience.
🐛 Bug Fixes
Agent Weaver – Infinite Loading After Concurrent Edit
Fixed an issue in Agent Weaver that caused an infinite loading state when a user tried to continue in view-only mode after a concurrent edit. This unblocks collaborative workflows and allows you to view agents without getting stuck.
Weaver Limit Event not Appearing in PostHog
Resolved a bug where the system was not displaying Weaver limit events in the PostHog analytics platform. This fix allows for proper tracking and analysis of these events, improving our ability to monitor system usage and user behavior.
Billing System Limit Not Enforced
We fixed a critical issue where the billing system failed to enforce user-defined spending limits in the staging environment. This protects you from unexpected costs by ensuring your defined spending limits are strictly enforced.
MCP & Alexa Integration Production URL Not Updating
Fixed a bug where the production URL in the MCP & Alexa integration would fail to update after an agent was deployed. This ensures reliable connections and prevents integration failures for your production agents.
Resolved several UI bugs that were identified in the Component Sidebar. This improves visual consistency and resolves usability issues when you are interacting with this part of the interface.
Credits System – Credits Consumed Without Usage
Resolved a bug where credits were being incorrectly deducted from user accounts even when no usage was reported. This fix protects you from unfair credit loss and ensures your balance is accurate.
Credits System – Free User Restrictions Not Properly Enforced
We addressed a problem where free-tier users could continue using premium features after exceeding their monthly credit limit. The restrictions are now properly enforced to maintain the integrity of our plans.
Sprint 40: May12 – May18, 2025
Changelog: May 19, 2025
This sprint introduces powerful new functionality, including cost tracing in the Inspect window and image input support for the Image Generator. We’ve also focused on improving the user experience with several UI and billing system enhancements. Additionally, we have resolved critical bugs affecting code components and data management to improve platform stability.
✨ New Features
Trace of Task Cost in Inspect
We’ve added a new tab in the Inspect window that displays both overall and broken-down cost tracing for completed workflow tasks. This enables you to view and analyze task costs directly after a workflow completes, providing greater visibility into resource consumption.
The Image Generator can now accept image attachments as input, similar to the Gen AI LLM component. It understands binary inputs and treats multiple images as an array, allowing you to use existing images to guide your generations and expanding the component’s creative possibilities.
🚀 Improvements
The attachment button in the chat interface has been relocated to the right side, between the character count and the send button, for improved usability and visual consistency.
To solve overflow issues in the builder interface, the Share, Test, and Deploy buttons have been converted to icon-only versions, ensuring all buttons are visible and accessible.
Standardized Keyboard Commands
Keyboard shortcuts for both Mac and Windows have been aligned with Google standards for a more familiar user experience. We also fixed an issue where CMD+A incorrectly triggered the “prettify” action.
Exclude Internal Emails from Seat Charges
The billing system has been updated to exclude email addresses ending in @smythos.com from being counted as billable seats, reducing costs for internal team usage.
🐛 Bug Fixes
Code Components – AggregateError Failure in All Components
Fixed a critical AggregateError related to the Axios library that caused all code components to fail when executed in the production environment. This resolves the issue and restores full functionality to all of your code components.
Data Pool/RAG – Deleted Books Still Accessible and Data Spaces Reappear After Deletion
We resolved a major data integrity issue where books and data spaces deleted from the Data Pool would remain accessible and reappear in the UI after an RAG Search. This fix ensures that data can now be permanently deleted as expected, resolving potential data privacy issues and eliminating user confusion.
Fixed an issue in the debug bar where two scroll bars would appear when inspecting a component with a long execution response. This makes navigation less cumbersome and improves the user experience when you are reviewing long responses.
Sprint 41 May 19 – May 25, 2025
Changelog: May 26, 2025
This sprint introduces a new Partner plan to support our growing ecosystem and makes global variables more accessible directly within the agent card. We’ve shipped a large number of improvements focused on UI clarity, security, and user feedback, including an audit of our access control lists. Key bug fixes address performance issues like slow login times and resolve critical permission and UI inconsistencies.
✨ New Features
Create Partner Tier Plan with Special URL
We have implemented a new “Partner” tier plan, which mirrors the STARTUP plan but includes 1,000 seats. It is accessible via a dedicated URL (app.smythos.com/partner) to enable our partners to easily add and train clients on their accounts.
Add Global Variables to Agent Card
You can now add and manage existing global variables directly from the agent card settings interface. This change improves agent design efficiency by making these variables more accessible without leaving the agent configuration screen.
🚀 Improvements
We’ve removed bold formatting from tooltip text for better readability and made the “title” field optional on the deployment screen for greater flexibility.
Invite Member Modal
The “Invite Member” modal now includes a description and conditional warnings about seat limitations and potential charges based on your plan, improving billing transparency.
Run Completion Messages
Clear confirmation messages will now appear after a workflow completes, indicating success or failure and reducing uncertainty.
Debug Inject Component UI
The debug inject component has received several UI updates, including button repositioning and styling changes, to improve its usability and visual appearance.
Info Icons and Misc Fixes
We’ve added various information icons and resolved other minor issues to enhance UI clarity.
ACL Audit
A comprehensive audit of our access control lists for view, edit, and admin permissions has been conducted to ensure system security and proper access management are in place.
🐛 Bug Fixes
Excessive Login Time to Homepage
We’ve addressed an issue that caused the initial login to the application to take more than 20 seconds. This fix significantly improves the login process, preventing potential user abandonment and creating a much better first impression.
Inspect Box Expansion Issue
We fixed a bug where expanding the inspect box would only add whitespace instead of resizing the contents. This makes the feature more efficient and ensures you can make better use of your screen space when examining data.
Global Variables Permissions Issue in Builder Page
We resolved a critical permissions issue where users without edit permissions could still modify global variables on the builder page. This fix ensures that access is correctly restricted, preventing unauthorized users from making changes to critical configuration settings.
We fixed an issue where multiple sidebar buttons would remain highlighted simultaneously, creating a confusing UI state. The active sidebar is now clearly indicated, making navigation more intuitive.
Regression on Weaver Chat Box
We’ve corrected a visual regression with the chat box in Weaver, ensuring it now displays correctly and restoring the expected user experience and usability of the interface.
Sprint 42 May 26 – June 1, 2025
Changelog: June 2, 2025
This week’s sprint introduces a highly requested feature: the ability to export chatbot logs. We’ve also shipped several key improvements to the user experience, including a new upgrade modal and UI fixes for the model list and agent settings. Finally, we’ve resolved a number of bugs affecting UI consistency, data display, and permissions to enhance platform stability.
✨ New Features
Export Chatbot Logs
You can now export the full chat history of an agent. This initial implementation enables you to access and analyze complete chatbot conversation histories, significantly improving your debugging and analytics capabilities. Future updates will include filtering options and other UI improvements.
🚀 Improvements
Partner Plan Invite Modal
The “invite member” warning modal will no longer appear for users on the new Partner plan, matching the seamless experience of our Enterprise plan customers.
Upgrade Modal on ‘My Plan’ Page
A new upgrade modal has been implemented on the “My Plan” page, following Figma design specifications to provide a clearer pathway for upgrading your subscription.
Model List Responsiveness
We’ve fixed layout issues with the model list by moving tags under the model name to better handle overflow, improving the list’s usability.
Agent Settings UI Fixes (Phase 3)
We have implemented another round of UI/UX fixes for the Agent Settings module to continue improving the interface and user experience.
🐛 Bug Fixes
Coupon UI Component Inconsistencies with MVP Mockup
We addressed multiple UI discrepancies in the coupon component, including incorrect buttons and text, to ensure it aligns with our approved design mockups. This improves design consistency and the overall user experience.
We fixed a bug where the entire chatbot background would incorrectly adopt the header’s background color. This fix restores proper contrast and improves readability and accessibility.
Builder Plan Shows Incorrect Data Pool Usage Limit
We corrected a discrepancy where the “My Plan” page for the Builder plan showed an incorrect data pool limit. This eliminates confusion and ensures the details on your plan page are accurate.
We fixed an issue where changes to a space’s name were not immediately reflected in the UI. This removes the need for a manual page refresh and makes the interface feel more responsive and intuitive.
Fix Agent Permission Modal Text Display
The agent permission modal was incorrectly displaying generic “[modify/view]” text. We’ve fixed this to show either “modify” or “view” based on your current access level, which improves clarity by showing the precise action you are permitted to take.
Analytics – Consumption Data Not Displaying in UI
We resolved an issue where the analytics page was not properly rendering model consumption data despite it being available in the API. This fix restores this important visualization, allowing you to accurately track your model usage for budgeting and planning.
For any questions or feedback, please reach out to our support channels.
