🎉 New Features​

📅 Scheduled Triggers​

Take your automation further by scheduling the execution of your agents or workflows with Serenity Star. Define the frequency, version, and values of the input parameters, and leave the work to us.

Note: Scheduled triggers require activation in your subscription plan. Contact [email protected] for early access.

📎 Microsoft Teams Integration​

We've added a new channel for your assistants: Microsoft Teams. Choose which agents will be available through your Teams bot on the Channels tab of the Agent Designer.

Note: To start using it, the Teams bot must be activated on your Azure subscription, contact us to complete this process.

🤖 Serena​

🌐 Create Agent from a Website​

Serena now uses web search directly to generate the base template for the agent. The base agent has been significantly improved with foundational skills and a more robust prompt.

🗂️ Serena Helps You Define Your Output JSON Format​

Serena is now available to help you when defining the output JSON format of your activity agent by providing helpful quick actions.

💡 Serena Helps You Define Your Company Guidelines​

When defining your company guidelines, you can use Serena to help you configure them with integrated Markdown rendering.

🛠️ Enhancements​

⚡ Simplified Agent Execution​

The conversational agent execution endpoint no longer requires creating a conversation first — if no ChatId is provided, a new conversation is created automatically. This streamlines one-off or programmatic executions while reducing unnecessary API calls.

We've also introduced an endpoint to retrieve initial agent context (welcome message, starters, etc.) without creating a conversation — ideal for pre-loading UI elements efficiently.

Relevant Endpoints​

• ▶️ Execute Agent: Use without chatId to auto-create a conversation: Link
• ℹ️ Get Initial Agent Conversation Info: Retrieve initial bot message and conversation starters without side effects: Link
• ⚠️ (Deprecated) Create Conversation: Still available temporarily but will be removed; migrate to the execution flow above: Link

📄 See API documentation for full request/response schemas.

📊 Agent Insights and Forms Endpoints​

Date filtering is now available for both Agent Insights and Agent Forms.

Relevant Endpoints​

• 📋 Get Forms by Agent: Retrieve forms data filtered by date: Link
• 📊 Get Insights by Agent: Retrieve insights data filtered by date: Link

📄 See API documentation for full request/response schemas.

🛡️ Improved Agent Restrictions​

You can now define the level at which you want to restrict your agent. This simplifies configuration and ensures that security restrictions are applied. We recommend always setting an audit-level restriction to protect user privacy. Link

🔑 Workflow Security​

A default API key is now created for workflows to ensure secure execution.
You can find it in Developer Tools.

🔐 Login Security​

Stronger security measures: the lockout process was improved and enforced for all users after 3 failed attempts.

🐞 Fixes & Improvements​

• Simplified Assistant Behavior definition in Agent Designer.
• Improved robustness in the WhatsApp channel.
• Layout and UI adjustments.
• Fixed Agent Insight automation to correctly handle non-UTC instances.
• Fixed deletion of knowledge files via the API.
• Updated default transcription model to gpt-4o transcribe.
• Fixed a bug that prevented selecting sub-models when choosing a featured model.
• Performance improvements in file download and client-side rendering.

🎉 New Features​

Workflows​

Unleash powerful automation with custom workflows. Seamlessly chain agents, trigger specific actions, and build end-to-end flows — all through an intuitive interface.

• Automate repetitive tasks: Reduce manual effort by letting workflows handle routine processes.
• Integrate multiple agents: Coordinate actions between different agents for complex scenarios.
• Conditional logic: Set up branching flows based on data, user input, or agent responses.
• Multi-step processes: Orchestrate sequences such as approvals, notifications, or data transformations.
• Trigger external actions: Connect with APIs, send emails, or update external systems as part of your workflow.

Note: Workflows require activation in your subscription plan.

Skills​

🧠 Code Execution Skill [Preview]​

Enable agents to generate and execute python code with the new Code Executor Skill.

🔁 Workflow Execution Skill​

Trigger custom workflows directly from agents to extend automation capabilities.

💡 Serena Skill Helper​

Leverage Serena during skill configuration to generate optimized descriptions and apply best practices — all with a single click.

🛠️ Enhancements​

💬 Serenity* AI Chat​

Navigate your agents, models, and settings more efficiently with our redesigned interface.

⭐ Featured Models​

Access a curated selection of high-performing models — now highlighted directly in the model selector for quick discovery.

🔄 Unified Agent Types​

All subscription plans agent features are now unified — combining Assistant, Activity, AI Proxy, and Chat Completion features into a single, powerful offering. Select from any of this options when creating a new agent

Track cost and usage more accurately when using your own API key.

🖥️ System Context​

Allow agents to access runtime system info with new templating options: Use the following expressions in your agent for instant access to advanced system data.

{{system.channel}}  
{{system.groupIdentifier}}  
{{system.knowledgeTags}}  
{{system.messages}}  
{{system.userIdentifier}}  
{{system.volatileKnowledgeIds}}  
{{system.whatsAppPhoneNumber}}  
{{system.stream}}  
{{system.date}}  
{{system.country}}

🔐 Login Security​

Stronger security measures: when registering a new login method, a confirmation email will now be sent for validation.

🌐 Web Search Skill​

Added support for Anthropic web search tools, unlocking powerful search capabilities for Claude models.

🐞 Fixes & Improvements​

• Improved naming conventions and visual styling in the Knowledge panel.
• Enhanced logout behavior: users are now correctly redirected to login.
• Agent Designer sliders now show token values with clear multipliers (e.g., 10K, 160K).
• Fixed “Insights on Demand” errors in Activity-type agents.
• Improved accuracy of date filters in Agent Manager and Cost Dashboard.
• Corrected filtering logic in the Dev Tools grid.
• Instance grid now preserves filter and UI state across sessions.
• Fixed “Update Now” redirect to correctly lead to the Change Plan screen.
• Resolved issue with in-memory cache and skill result reuse.
• Reinstated missing cancel button in the duplicate agent side panel.
• Prevented duplicate submissions from multiple rapid clicks on the submit button.
• Added retry logic and advanced logging for WhatsApp method to improve error detection.
• Fixed code snippet rendering in the agent demo.
• Enhanced performance and token formatting in agent execution views.
• Agent grid now displays active filters clearly.

🎉 New Features​

🤖 Agent Prompts powered by Serena​

Quickly improve your prompts using Serena.
Click the Serena icon to add custom instructions to your prompts. You can also choose any of our "Quick actions" to enhance your prompt with predefined instructions.

🧩 Skill Templates​

Create and share custom skill templates across your organization for consistency and reuse. Save time by leveraging preconfigured templates.

Then, you'll be asked if you want to start from a template or create a new skill from scratch.

💬 Serenity* AI Chat​

🏢 Subtenancy Support​

Serenity* AI Chat now supports Subtenancy and Agent Permissions.
Users have exclusive access to Agents within their own Subtenant, improving compartmentalization and security.

🔗 Share Conversations​

Create conversation snapshots to quickly share ideas with anyone.

You can revoke access to shared conversations at any time, ensuring control over your data.

🛠️ Enhancements​

Agent Instance​

Access volatile knowledge files attached to user messages directly from the conversation history for better analysis.

📝 Genie Text Component​

A new assisted mode for the Genie Text Area lets users send custom instructions to the AI for improved response quality.

🐞 Bugs & Improvements​

• Fixed error when editing a Subtenant.
• Forms can now be inactivated for better control during agent execution.
• Improved visualization of Skill cards when descriptions are lengthy.
• Volatile knowledge files are now visible in agent instances.
• Fixed bug caused by double-clicking the edit button on a skill.
• Upgraded Skill emulator to GPT 4.1 Mini.
• Added more evaluator models for improved assessment.
• Enhanced performance with better validations during Agent Execution.
• General UI adjustments for improved usability.
• Corrected arrows and tooltips in agent and cost dashboard totals.
• Fixed multiple submit errors in the AI Chat configuration screen.

🎉 New Features​

🔄 Fallback Model​

Ensure continuous operation by selecting a fallback model. If your main model fails due to a vendor issue, the fallback model will be used automatically.

Requires the Fallback model feature in your subscription plan.

🔗 Export Agents​

Generate a public link to share your AI agent and its configuration. Anyone with the link can copy and reuse your agent—ideal for collaboration, showcasing, or cloning.

No sensitive information (such as authorization keys) is included in the copied agent.

🔑 Bring Your Own Key​

Manage and use your own AI model provider keys. Executions using your key do not generate charges on your Serenity balance.

Requires the Bring Your Own Key feature in your subscription plan.

✨ Genie Textarea​

Introducing a powerful AI-enhanced textarea web component for Serenity* Star. Instantly transform any textarea into an intelligent editor with AI-powered completions and content generation.

Acess the documentation here

⚙️ Enhancements​

🛡️ Alinia Guardrails​

Integrate your own Alinia account with Serenity to access advanced guardrails features.

🛠️ Skills​

🔄 Reuse Skill Output​

Agents can now reuse information from previous skill executions for faster responses.

🌐 HTTP Request Skill​

Header responses can now be included in the agent context.

Newly supported content types:

• application/json-patch+json
• application/merge-patch+json
• application/ld+json
• application/hal+json

Default Accept headers added:

• application/vnd.api+json
• application/json-patch+json
• application/merge-patch+json
• application/ld+json
• application/hal+json
• application/problem+json
• application/geo+json
• application/schema+json

🤖 AI-Powered Knowledge​

The new knowledge integration mode intelligently filters tags and optimizes user queries to deliver more relevant responses. The agent now leverages chat history to improve result relevance.

🧠 Model Selector​

The improved model selector lets you choose from over 300 LLMs, easily selecting your preferred vendor or snapshot.

📊 Insights​

Trigger evaluation of insights on demand for a specific conversation or agent instance.

🐞 Bugs & Improvements​

• Added current subscription billing cycle information to billing view.
• Input parameters from preview are now automatically copied to code snippets.
• Improved access to insights grids when the feature was not available.
• Fixed bug in agent behavior and definition with advanced commands.
• Corrected typos and improved localizations.
• Improved insights processing job for better coverage and enhanced agent analysis.
• Avoided resetting Knowledge tags when adding a new file.
• Fixed parameter handling when executing the skill multiple times.
• Fixed duplicate agent error when the original agent had permissions.
• Fixed error that prevented uploading more than one knowledge file at a time.
• Fixed error when changing embedding model in agent.
• Fixed AI Chat Configuration activation of speech generation skill.
• Added default knowledge integration mode in AI Proxy Agents.
• Fixed insights mapping when duplicating agents.
• Automatically removed free plan quota when upgrading to paid plan.
• Fixed bug in chat widget with agents that had vision activated.

✨ New Features​

🤖 Agent Insights​

Agent Insights allow you to automatically analyze past conversations and evaluate specific aspects of agent interactions.

You can define what insights to collect and set the desired coverage. Conversations will then be periodically analyzed to extract the defined data.

Requires the Agent Insights feature in your subscription plan

Agent Insights analysis incurs a processing cost

🎯 Conversation Objective​

Assistant and Copilot agents now support setting a Conversation Objective, which automatically closes the conversation when the specified goal is met. This helps reduce costs and avoid excessive use of a single conversation.

🚀 Performance Improvements​

🗄️ Cache Optimization​

We’ve reevaluated and improved cache integration, resulting in smoother operation across all hub layers.

📚 Knowledge​

🧠 Extended Context Knowledge​

The new knowledge integration mode improves how information is organized and retrieved thanks to expanded context windows.

Enable Large Context Knowledge in the advanced settings to access this new behaviour.

Improvements include:

• Grouped file chunks for better file organization
• Enhanced metadata for richer context
• Refined similarity search with extended conversation history analysis for higher knowledge relevance

🏷️ Custom File Metadata​

You can now define custom metadata for each knowledge file to help the agent assess relevance during conversations.

Files can be filtered using tags by including the knowledgeTags parameter in agent executions:

curl -X POST \
 -L "https://api.serenitystar.ai/api/v2/agent/<AGENT_CODE>/execute" \
 -H "Content-Type:application/json" \
 -H "X-API-KEY: <YOUR_API_KEY>" \
 -d '[ 
  ...
  {
    "key": "knowledgeTags",
    "value": "Tag1, Tag2, Tag3"
  },
  ...
]'

Knowledge tags work by applying OR logic

🛠️ Enhancements​

🔑 API Standardization of Task Messages​

Server-Sent Event task messages are now standardized using specific keys:

{ LoadingConfiguration, "configuration_load" },
{ ValidatingConfiguration, "configuration_validate" },
{ ValidatingRequest, "request_validate" },
{ AppendingContext, "context_append" },
{ CalculatingTokenBudget, "context_token_budget_calculate" },
{ SearchingRelevantKnowledge, "context_knowledge_search" },
{ SearchingRelevantVolatileKnowledge, "context_volatileknowledge_search" },
{ AskSensibleDataObfuscation, "sensitive_data_system_definition_obfuscate" },
{ InputSensibleDataObfuscation, "sensitive_data_user_input_obfuscate" },
{ CreatingInstance, "agent_instance_create" },
{ UpdatingActivity, "agent_instance_update" },
{ UpdatingChat, "agent_instance_update" },
{ UpdatingProxyRequest, "agent_instance_update" },
{ CalculatingPrices, "agent_instance_fares_create" },
{ ExecutingMetacontrol, "safety_metacontrol_execute" },
{ ExecutingAlinia, "safety_alinia_execute" },
{ InputModeration, "safety_input_moderate" },
{ LoadingPlugins, "skills_load" },
{ ExecutingPlugin, "skills_{0}_execute" },
{ InvokingPluginsWithEmulator, "skills_emulator_execute" },
{ ValidatingOutputJsonFormat, "output_format_validate" },
{ GettingAIResponse, "llm_response" }

🧩 Skills: Active/Inactive Toggle​

Skills can now be toggled as active or inactive to control whether they're included in agent execution. This allows you to experiment without losing skill configuration.

📝 Markdown Editor in All Agents​

All agent types now support a built-in markdown editor. Markdown rendering performance and the output JSON field display size have been improved.

🧘 Serenity AI Chat​

Voice interaction​

Use voice commands to send messages to your conversation.

Automatic Guidelines​

When registering with a corporate email, AI Guidelines are now automatically generated based on your company domain.

🐞 Bugs & Minor Improvements​

• Exported instance cost added to Quality Studio
• Improved validation in system definition and behavior
• Fixed logging error during agent designer validation
• Excel export in Agent Manager now differentiates input and output tokens
• Added Forms API endpoint
• Fixed user usage chart in AI Chat Management
• Resolved restore issue in chat widget when agent sent audio
• Added support for application/vnd.api+json header in HTTP request skill
• Fixed demo page rendering for agents with parameters
• Fixed skill configuration issues when user had agent permissions but lacked required roles
• AI Chat Guidelines now generated from email domain on tenant registration
• Improved isolation of Chat Widget styles
• Added endpoints to retrieve Forms results and Agent Insight results
• Fixed executor skill issues when the invoked agent had no input parameters
• Fixed side panel rendering in agent designer for newly added users with permissions
• Improved scroll behaviour on Serenity* AI Chat
• Improved error handling on Serenity* AI Chat
• Fixed bug in edit message on Serenity* AI Chat

🚀 New Features​

👥 User Access​

Define who can edit, audit, or execute your agent with full customization. Enhance security by providing custom permissions for each user.

Administration: Grants the user administrator privileges on the agent, allowing them to edit and modify the agent.

Audit: Grants permission to the user to view the conversation history of the agent, and access costs and usage dashboards.

Execution: Grants permission to execute the agent.

📝 Markdown Editor​

Assistant agents now have an integrated markdown editor to better visualize and define the behavior of the agent.

Full screen support enabled

Enhancements​

💬 Serenity* AI Chat​

⚙️ Detailed Management​

Serenity* AI Chat configuration is now Serenity* AI Chat management. Configure and access all relevant information about Serenity* AI Chat usage in a single view.

Access skill configuration, view conversation history, and track usage.

Configure custom skills to be used in your AI Chat conversations for better interactions

Access the full list of conversations directly from Serenity* AI Chat Management

Easily track usage and cost statistics

🔄 Usability Improvements​

Serenity* AI Chat has received major usability improvements, allowing for concurrent conversations and better stability when managing conversations.

Improved navigation options allow for message navigation history using CTRL + ⇅.

📊 Agent and Cost Management​

User Consumption​

Track exactly how each user is using Serenity* Star and monitor input/output token distribution.

Usage by user is tracked based on the 'userIdentifier' parameter in the agent execution, allowing for client tracking of usage events.

Token usage can be filtered by input or output tokens.

Remaining Token Balance​

Easily view your use of included tokens in your current subscription plan in your Cost Dashboard.

🧪 Agent Quality Studio​

Export Test Plans​

Test plans can now be fully exported for customization in an Excel file. When importing a test plan, if matching test case codes are found, test case information is updated based on the imported files.

Reestablish Parameters from Global Parameters​

All agent types now have the option to declare global parameters. When declaring individual test case parameters, they can easily be reestablished based on global parameters for quicker configuration.

Performance​

Major improvements have been made in performance across all layers of the project. You can expect:

• Faster API responses
• Smoother and optimized web app navigation

Input Parameters​

Default values declared for input parameters in the agent designer are used in agent execution if no matching key is found in the agent execution.

Added more robust and clearer validations for type mismatches.

Bugs & Improvements​

• Improved redirection to agent instance from agent designer preview when a vendor error occurred
• Improved sensitive data obfuscation functionality in all agent types
• Fixed issue on agent duplication where JSON response option format was lost
• Fixed issue in API key view button on dev tools
• Improved styles and validation errors in various views
• Improved tax ID visualization in billing preferences
• Fixed rendering of JSON format radio button on agent designer when duplicating agents
• When a model does not support skills, an information box is displayed on agent designer
• Fixed API Key visualization on Dev Tools
• Improved markdown table support on agent instance details
• Added support for 2-Factor Authentication
• User lockout is restored when recovering password
• API returns execution cost
• Added currency symbol to execution limits
• Improved auto-scroll on compact mode in Chat widget
• Set audio recording from Chat Widget channel

🎉 New Features​

🔑 Agent Permissions​

Get full control over agent security by creating specific API keys for each agent.
New scoped keys can be easily created and managed from the Agent Designer view. This provides full flexibility on agent usage and security.

Available permissions:

• Administration: Can perform any operation related to the agent on the API, including execution and auditing.
• Audit: Can access the list of conversations, specific instances, and usages of the agent on the API, including agent execution.
• Execute: General permission for only executing an agent and accessing a given conversation.

⚙️ Enhancements​

🔄 Change Agent Type​

When duplicating an agent, you can choose to change to a new agent type. This maintains all matching definitions in the new agent.

💬 Serenity* AI Chat​

📍 Grounding Generation With AI​

Easily generate custom guidelines for AI Chat by completing a form and letting Serena create an advanced prompt.

🔒 Security​

Enhance AI Chat security by supervising interactions and setting usage limits.

🤖 Agent Designer​

▶️ Try It Out​

Easily try your agent from a demo page or from AI Chat by accessing the Try It Out feature.

Channels are automatically activated so that you can enjoy the agent as fast as possible.

📝 Access Agent Instance​

Quickly access conversation details through the new instance button in Agent Designer's Preview section, providing insights on your current interaction.

🧠 Skill Support in Proxy Agents​

Proxy agent requests can now include instructions for executing skills.

Check out the full usage instructions in the Proxy Agent documentation

🧹 Conversation Cleanup​

Conversations with no registered user messages are deleted automatically after 24 hours, avoiding incorrect usage statistics.

🤖 New Providers​

• Mercury Inception Labs

🐛 Bugs & Minor Improvements​

• Cleanup of unused conversations after 24 hours.
• Show markdown in system agent instance responses.
• Added new user role Agent Usage with access to Agent Manager and Cost Dashboards in agents.
• Fixed import of test cases.
• Allow activating or deactivating pay-as-you-go business model.
• Added Tax ID validation to avoid duplicate values.
• Styles and UI improvements in various screens and agent instances.
• Improved validation of required parameters in Agent Quality Studio when creating Test Plans and Test Executions.
• Added guardrails support in Activity, Chat, and Proxy agents.
• Forms return completion data on API when form is completed.
• Improved usability in the Tax ID component.
• New agent usage role for accessing cost and management dashboards.
• Added vision support for QWEN 2.5 VL from Alibaba.
• Improved card rendering of guardrails card in Agent Designer.

🎉 New Features​

🤖 Serenity* AI Chat - Company Grounding​

📚 Company Knowledge​

Boost AI Chat responses by uploading custom knowledge files. These are automatically accessible in every AI conversation.

🧭 Company Guidelines​

Define high-level rules and usage parameters for Serenity AI Chat:

• Set conversation guidelines
• Limit or define allowed topics
• Customize agent behavior globally

🤖 Curated List of Models​

Define if you wish to use your own list of models or use a curated list of models by the Serenity* Star team.

⚙️ Enhancements​

💳 Billing​

Configure and manage your billing preferences to ensure compliant invoice generation.

• Enable business invoice settings
• Declare tax IDs
• Set and manage billing address
• Collected taxes are validated against the customer

New invoices now include this data to support taxability compliance.

📝 Forms​

Collected form data is now visible directly within the Agent Instance View, improving data access and clarity.

🧪 Agent Quality Studio​

📥 Import Test Plans​

You can now import test plans directly from an Excel file:

1. Create your plan using our downloadable template.
2. Upload the file.
3. All test cases are imported and linked to the current agent version.

New Evaluator Models​

New evaluator models have been added to Agent Quality Studio for better results

• GPT 4.1 Mini
• GPT 4.1
• Qwen Max
• Claude 3.7 Sonnet

🔐 Security – API Keys​

To reduce security exposure:

• A dedicated API key for agent execution is now generated per tenant.
• Prevents main keys from being exposed to external clients.
• The new key is automatically included in code snippet examples.

Bugs & Minor Improvements​

• Show rate limits features on current plan card
• Removed close conversation modal on Activity Agent version grid when publishing versions
• Agent Executor skill only allows for agents in the same subtenant to be executed
• Fixed agent execution skill when using subtenant key
• Add audit footer to system agent instances
• Validate category names in AI Chat configuration
• Fixed bug in Chat Widget when conversation was closed and no messages where sent
• Fixed create trial failing
• Typos and text improvements
• When updating AI Chat configurations user is redirected to home
• Do not show +/- icons on features that cannot be expanded
• Fixed visualization of Serenity* AI Chat Channel in Agents when user limit is met
• Fixed prompt generation of Serena in System agent when no matching template was found

🎉 New Features​

📋 Forms​

Enable a new way for agents to interact and collect structured data directly from the conversation. Forms allow your agent to generate tables with predefined fields — perfect for collecting leads, registering activities, or any other custom workflow.

• Create and manage custom forms
• Define exactly when and how the agent should gather data
• View submissions in a clean, structured format

Streamline your operations with real-time, conversational data collection.

🌐 Web Search Skill​

Expand your agent’s capabilities with the Web Search Skill. Agents can now retrieve up-to-date information from the internet in real time.

Deliver more accurate, relevant responses — even when dealing with the latest news and facts.

🛠️ Enhancements​

🤖 Serenity* AI Chat​

🖼️ Vision Support​

Serenity* AI Chat now supports image processing when the selected model includes vision capabilities. Give your agent visual context to enhance understanding and response quality.

⚙️ Custom Models & Categories​

Gain full control over which models are available in Serenity* AI Chat. Define default and available models and create categories to organize model selection.

Improve usability and align model access with your operational needs.

👤 Quick Access to Profile​

Access your Serenity* AI Hub account directly from the chat interface — fast and seamless.

💳 Extra Credit for Payment Method​

Receive an additional €5 in credits when you add your first payment method to Serenity* AI Hub. A small bonus to get you started with more freedom.

⚠️ Plan Agents Deprecated​

Plan Agents have reached end of life and are no longer supported. All functionality can now be replaced using Activity Agents, which offer a more powerful and flexible approach. Upgrade to a more robust agent model without losing functionality.

🧪 Agent Quality Studio​

📄 Duplicate Test Cases​

You can now easily duplicate test cases in the Agent Quality Studio, making test plan creation faster and more efficient.

📚 API Knowledge​

A new API endpoint is available to upload knowledge files using multipart/form-data. Simplify integration and keep your agents well-informed with up-to-date resources.

curl -X 'POST' \
  'https://api.serenitystar.ai/api/KnowledgeFile/upload/<AGENT_CODE>' \
  -H 'accept: */*' \
  -H 'X-API-KEY: <YOUR_API_KEY>' \
  -H 'Content-Type: multipart/form-data' \
  -F 'ContentType=' \
  -F 'ContentDisposition=' \
  -F 'Headers={
  "additionalProp1": [
    "string"
  ],
  "additionalProp2": [
    "string"
  ],
  "additionalProp3": [
    "string"
  ]
}' \
  -F 'Length=' \
  -F 'Name=' \
  -F 'FileName='

🤖 New Model Providers​

• Nextbit 256

🐞 Bug Fixes & Improvements​

• Restored ability to recover deleted social login accounts.
• Added validation for required user identifier parameters in test plans and executions.
• Improved metadata handling for Volatile Knowledge.
• Fixed issues with transcription and demo page audio recordings.
• Suggested top-up and auto-recharge amounts are now displayed when balance is low.
• Balance changes and subscription plan updates are now emailed both to the user and the contact address.
• Fixed skill execution errors outside of preview mode.
• Improved agent executor performance in message field calls.
• Expanded skill support for Anthropic models.

🎉 New Features​

End Conversation​

Enhance control over conversation lifetimes with the End Conversation feature. Conversations with an agent can now be automatically closed based on predefined criteria, ensuring better session management and enforcing limits such as inactivity periods or maximum duration.

Reasons for Conversation Termination:​

• Maximum number of user messages reached
• Quota limit exceeded
• Inactivity period exceeded
• Time limit reached
• Manually closed by an administrator
• A new version of the agent is published

Conversations now include an End Date, and their closure can be configured in the Security Tab. All new agents will have a default time limit of 24 hours:

Once a conversation is ended, the user will no longer be able to continue it:

The Conversations Grid now displays the status of conversations and their respective end reasons:

Administrators can manually finalize a conversation at any time:

Additionally, when publishing a new agent version, all active conversations can be forcibly closed to ensure all users interact with the latest version:

Chat Widget Integration​

The chat widget now supports the End Conversation feature, allowing administrators to define a message that users see when a conversation ends:

On the WhatsApp channel, a new conversation will automatically start when the previous one is closed.

Improved Rate Limits and Quota Usages​

Enjoy better rate limits in the API by upgrading your subscription plan.

🛠️ Enhancements​

Standardized Error Messages​

Error messages during agent execution have been standardized, improving API response clarity. Below are the updated error codes:

• message_required: An input message was not sent to the conversational agent.
• conversation_closed: The current conversation is closed; start a new one.
• parameter_type_mismatch: Provided parameters do not match agent specifications.
• insufficient_balance: Insufficient balance; add credit to continue.
• missing_group_identifier: The groupIdentifier parameter is required.
• required_parameters_missing: A required parameter is missing in the request.
• required_parameters_null: A required parameter was sent with a null value.
• required_parameters_empty: A required parameter was sent with an empty value.
• monthly_quota_exceeded: The monthly quota for the agent has been exceeded.
• model_quota_exceeded: The selected model quota has been exceeded; choose another model.
• organization_quota_exceeded: Organization-wide quota exceeded; increase limits.
• user_quota_exceeded: User quota exceeded; increase user limits.
• missing_user_identifier: A required user identifier was not provided.
• unfinished_volatile_knowledge: The requested volatile knowledge file is still processing.

HTTP Request Skill​

• A User-Agent header is now included in all API requests.
• Error codes and stack traces are now displayed when a request fails:
  

Transcribe Service​

• The default transcription model has been updated to GPT-4o Mini Transcribe, improving accuracy and efficiency.

Billing Preferences​

• Users can now customize their billing and notification contact information:
  

• Configure low balance notifications based on individual requirements:
  

Security​

Moderations​

• All free-tier user requests now pass through OpenAI’s moderation endpoint to ensure compliance and security:
  

Alinia Integration​

Integration with Alinia has been added for improved security and performance. Activate safe use of LLM for guardrails against:

• Investment
• Legal
• Medical
• Tax

🤖 New Models​

New models have been introduced, enhancing AI capabilities and performance.

• Open AI GPT 4o Mini TTS
• Gemini 2.5 Pro Experimental
• Alibaba Qwen 2.5
• Alibaba Qwen Turbo
• Alibaba Qwen Plus

🐞 Bug Fixes & Improvements​

• Improved performance of volatile knowledge processing
• Increased max description field size in all skills (from 1,000 to 4,000 characters)
• Fixed array parsing of input parameters in Agent Quality Studio
• Resolved issues in the demo page** when non-required parameters were omitted
• Added sender name in the Email plugin
• Avoid duplicate email sending in Email plugin if a email is repeated
• Fix bug when using the required user identifier and executing a realtime model on the demo page
• Fix user identifier validation on Test Plan and Test Execution
• Fixed mapping of arrays in input parameters when using Agent Quality Studio
• Corrected time plugin creation.
• Added name length validation on api key creation.