Conversation Overview
The Conversation Log is accessible within a selected session. It provides a complete history of messages exchanged between the user and the chatbot, including additional details such as function calls, token usage, and response time. This data helps you monitor how effectively the chatbot handles conversations and identify areas for improvement.
Key Features of the Conversation Log
1. Message Flow
Messages are displayed in the order they were sent and received, allowing you to follow the conversation step-by-step. Each message includes:
| Title | Description |
|---|---|
| Sender | Indicates whether the message was sent by the user or the chatbot. |
| Message Content | The actual text of the message exchanged. |
| Timestamp | The date and time when the message was sent. |
This provides context and helps diagnose any communication issues during the conversation.
2. Function Calls (for Sent Messages)
When a chatbot message triggers a function call, such as retrieving data or sending an email, the following details are displayed:
| Title | Description |
|---|---|
| Function Name | The name of the function executed (e.g., GetWeather, SendEmail). |
| Input Parameters | Details of the parameters sent to the function (e.g., location, email address). |
| Response Data | The output returned by the function (e.g., weather forecast, success message). |
| Status | Indicates whether the function call was successful or encountered an error. |
Function calls help extend the chatbot’s capabilities by integrating external services and APIs.
3. Token Cost
Each message consumes tokens, which measure the amount of data processed by the chatbot. Token usage is displayed for sent messages, making it easy to track how much your chatbot is using within a session.
| Metric | Description |
|---|---|
| Tokens Used | Number of tokens consumed for the message (including prompt and response). |
| Token Limit Impact | Helps determine if token usage needs to be optimized to stay within subscription limits. |
Efficient token management ensures better performance and adherence to subscription quotas.
4. Response Time
The time it takes for the chatbot to respond is measured and displayed for each message. Long response times can indicate performance issues or heavy processing tasks.
| Title | Description |
|---|---|
| Response Time | The time taken for the chatbot to process and send a response. |
Monitoring response times helps optimize system performance and maintain user engagement.
How to Use the Conversation Log
1. Analyze Message Patterns
Review the full conversation history to understand user queries and chatbot responses. Identify if there were any misunderstandings, delays, or irrelevant responses.
2. Review Function Execution
Ensure that function calls triggered by the chatbot (e.g., fetching data or sending a message) are working correctly. Check if the input parameters and responses match expectations.
3. Optimize Token Usage
If a session has high token usage, you can analyze individual messages to identify where improvements can be made, such as reducing prompt complexity or shortening responses.
4. Track Performance
Monitor response times to identify potential performance bottlenecks. If response times are consistently high, consider optimizing backend processes.
Use Cases
-
Debugging and Troubleshooting
Use the conversation log to diagnose issues with chatbot responses, including errors in function calls or delays in processing. -
Performance Optimization
Analyze token usage and response times to optimize system efficiency. -
User Experience Improvements
Identify patterns in user queries and feedback to improve the chatbot's understanding and responses.
FAQs
-
What is a function call in the conversation log?
A function call is a request made by the chatbot to an external service, such as fetching data from an API. It includes details like input parameters, response data, and status. -
How can I reduce token usage?
Review and optimize your prompts and responses to be more concise while still delivering the necessary information. -
What should I do if response times are too high?
Check for backend performance issues, such as delays in function execution or heavy processing tasks. Optimize these areas to improve response times. -
Can I export the conversation log?
Yes, the conversation log can be exported as a CSV file for further analysis.
The Conversation Log provides essential insights into the flow of communication, function calls, and performance metrics, helping teams monitor and improve chatbot interactions.