Release 4.1 2024.9.23.1
π New featuresβ
Subtenancyβ
The new Subtenancy feature in Serenity AI Hub allows organizations to create subtenants within their account, providing greater control and data separation. Subtenants operate as independent entities, managing their own agents without mixing data with other subtenants or the main organization.
Agents can now be assigned to either a subtenant or the root tenant.
Key points:
- Agent Ownership: Agents can be assigned to a specific subtenant or the root tenant, ensuring clear ownership structures.
- Data Isolation: Users from different subtenants are restricted from accessing each otherβs data or agents, maintaining strict boundaries between entities.
- Security: The new structure improves security by isolating data between subtenants, preventing unauthorized access.
- User Management: Users can now be assigned to one or more subtenants, offering flexible control over organizational access.
Subtenants can be managed at Settings -> Subtenant
Requires Subtenancy as Feature in your subscription
Agents are filtered by subtenant, use the funnel to impersonate the logged user in a given subtenant.
Agent manager overviewβ
Requires agent manager feature in your subscription
The Agent Manager now includes advanced statistical insights to help you monitor and optimize agent usage more effectively. This new view offers a comprehensive breakdown of agent performance and resource consumption, providing key data points for better decision-making.
Key Insights Available:
- Total Instances: View the total number of agent instances.
- Active Agents: Track how many agents are currently active.
- Avg. Tokens per Instance: Analyze the average number of tokens used per instance.
- Total Tokens Used: Get a complete overview of the total tokens consumed.
- Agent Type Distribution: See how agents are distributed by type.
- Model Usage Distribution: Understand which models are being used across different agents.
This detailed overview allows for better resource management and optimization of agent performance.
Agent Securityβ
To strengthen agent security, execution limits can now be set to manage resource consumption more effectively. Administrators can define quotas and rate limits, ensuring agents operate within defined parameters and reducing the risk of overuse.
Key Management Options:
- Conversation Quotas: Set limits on agent usage per conversation.
- Monthly Usage Quotas: Define monthly limits for agent interactions.
- Rate Limits: Control the frequency of agent requests over a specific period.
These controls provide greater flexibility in managing agent resources, ensuring optimal performance while maintaining security and preventing misuse.
Subscriptionsβ
Customers now have full control over their subscriptions within Serenity, allowing for effortless plan changes and billing adjustments. This new feature empowers users to manage their accounts without needing external support.
Key Features:β
- Subscription Management: Seamlessly switch between different plans at any time.
- Billing Overview: Access a detailed summary of your current billing, providing full transparency.
- Payment Methods: Easily update and manage your payment methods.
- Configuration Control: Adjust key subscription and account settings to fit your needs.
Billing Cycle Adjustments:β
- Pro-rated Discounts: When upgrading plans mid-billing cycle, Serenity recognizes any unused subscription and applies a discount based on the remaining time in your current billing period.
- Instant Credits: If downgrading, receive instant credit for any unused portion of your previous plan, ensuring fair and flexible billing adjustments.
This self-service feature simplifies subscription management, offering autonomy over your billing and plan configuration while providing flexible, customer-friendly billing adjustments.
π Enhancementsβ
Model Catalog - Advanced Search & Detailed Viewβ
The Model Catalog now offers enhanced search capabilities and detailed insights for easier model selection and management. This update helps streamline your workflow by providing more control over filtering and accessing model information.
Key Enhancements:β
Advanced Search Filters: Search models by key attributes such as:
- Size: Filter models based on their size to find the most suitable option for your use case.
- Language: Search models that support specific languages for better localization.
- Recommended: Easily identify models recommended for your particular needs.
- Detailed View: Gain deeper insights into each model with comprehensive information available in the details view, ensuring you make informed decisions when selecting models for deployment.
Real-Time Serenity Chat Customization - Agent Demo Pageβ
Serenity now offers real-time chat customization directly on the Agent Demo page, allowing you to quickly update, test, and modify the appearance and behavior of the chat interface. This feature provides full flexibility over the look and feel of your chat interactions without any downtime.
Key Features:β
- Instant Updates: Make changes to your chat's design and settings in real-time, with no need for redeployment.
- Visual Customization: Modify elements such as message colors and chat window appearance to align with your brand or specific use case.
- Advanced Parameter: Customize deeper settings such as theming, error messages and call to actions giving you full control over how your chat agents perform.
This enhancement streamlines the design and testing process, ensuring faster iteration and a personalized user experience.
Welcome email when registeringβ
New users registering with Serenity AI Hub will now receive a welcome email that offers a streamlined introduction to the platform. This feature ensures that new users can quickly get up to speed and start using the platform effectively.
Serenaβ
The Serena Agent Creator has been significantly updated to streamline the generation of AI agents. This enhancement focuses on system definition, template adaptation, and personality integration, allowing users to create high-performing agents more efficiently.
Key Enhancements:β
- Personality Integration: Seamlessly integrate specific personality traits into agents, enhancing user interactions and engagement.
- Advanced Prompt Generation: Benefit from more specific and detailed prompts that minimize the need for extensive prompt engineering, allowing you to focus on defining agents rather than crafting prompts.
Knowledge Chunkingβ
Introducing Knowledge Chunking, a new feature that allows users to control how their knowledge files are segmented. By utilizing the specified delimiter:
%%%%%%%----////////\\\\\\\\----%%%%%%%
users can define distinct sections within their knowledge files.
Key Benefits:
-
Enhanced Context Management: This feature provides greater control over how information is organized, ensuring that agents maintain context and relevance during interactions.
-
Precision in Chunking: Define exactly where the knowledge file should be chunked, allowing the agent to access critical information without missing important details.
-
Improved Knowledge Retrieval: By segmenting knowledge effectively, agents can retrieve and utilize information more efficiently, leading to enhanced performance and user satisfaction.
This enhancement empowers users to optimize their knowledge management processes, resulting in better-informed and more capable AI agents.
This will generate a new section within the knowledge file.
π Bug fixingβ
- Added Skill support for Mistral, Gemini.
- Time in Serenity* Chat is showed only on User Message.
- Reprocess of web pages takes most recenty version.
- Better handling of submit of side panels avoiding multiple submits.
- Solved authenthication issues when using the Demo Page and store conversation functionality.
- Fixed duplicate jobs being run.