Oct 7, 2024
This September, Smartly.AI enhanced transparency by improving observability in conversational agents, allowing you to trace interactions and understand decisions for better explainability.
This September, we've rolled out key updates to Smartly.AI that focus on improving transparency in conversational agents and strengthening platform security. Here's how these enhancements benefit you:
Improved Observability for Greater Transparency
Understanding how your conversational agent processes and responds to user inputs is crucial. With our enhanced observability features, you can now:
Trace Conversation Flows: Follow each step of the interaction, from the initial user input to the final response, making it easier to understand how decisions are made.
Input and Output Translation: Automatically translate user messages into the bot's language and responses back into the user's language. This ensures smoother communication, even if the user makes typos or errors.
Clearer Instructions: We've improved the readability of instructions within the observability panel, so you can quickly grasp how the agent operates.
These improvements provide:
Transparency: Gain full visibility into the agent's processes, building trust with your users.
Explainability: Easily explain how the agent works, which is essential for compliance and user confidence.
Efficient Debugging: Quickly identify and resolve issues within the conversation flow.
Strengthened Security with Rate Limiting
To protect against abuse and brute-force attacks, we've implemented a rate limit of 50 requests per minute per user for interactions via Webchat and the /dialog
API. This measure helps:
Prevent Abuse: Stops malicious actors from overwhelming the system with excessive requests.
Enhance Security: Shields the platform from unauthorized access attempts and potential exploits.
Ensure Reliability: Maintains consistent performance for all users by preventing service disruptions.
The detailed changelog for Sept 2024 is available here.
Looking Ahead: Exciting Features Coming in October
We're excited to announce that in October, we'll be introducing new features to further enhance your experience with Smartly.AI:
Evaluations: Utilize Large Language Models (LLMs) as judges to evaluate your bot's responses and track regressions. This will help you maintain high-quality interactions and identify areas for improvement.
Tools: Empower your agents with prebuilt or custom tools, allowing them to execute code, call APIs, send emails, and more. This expands the capabilities of your conversational agents, enabling them to perform complex tasks seamlessly.