How to Set AI Confidence Thresholds in Fullview
Fullview’s AI Confidence Score feature helps you control the reliability of responses generated by our AI agent. By adjusting the confidence threshold, you can ensure that the AI only responds when it’s sufficiently confident in its answers.
This article explains how to configure the AI confidence threshold in your Fullview account.
What Is an AI Confidence Score?
Fullview's AI evaluates every scenario and assigns a confidence score to each. This score indicates how confident the AI is in its response. The higher the score, the more reliable the answer is likely to be.
Why Set a Confidence Threshold?
As an admin, setting a confidence threshold determines when the AI is allowed to respond:
Above the threshold: The AI agent will answer the question.
Below the threshold: The AI will skip the question to avoid providing potentially unreliable information if training data is lacking. You can configure Fullview AI to route to your team in this case.
This ensures your users only receive high-quality, trustworthy responses from the AI. You can be notified when support scenarios fall under the threshold so you can apply more training data.
How to Set the Confidence Threshold
Log in to Fullview with admin credentials.
Navigate to Settings.
Click on the AI Agent tab at the top.
Use the slider to adjust your desired confidence threshold.
A recommended range is provided for optimal balance between automation and accuracy.
Setting the threshold above the recommended range may reduce the number AI responses, leading to potentially lower levels of overall automation and cost savings.
Best Practices
Start with the recommended range to maintain a good balance of coverage and confidence.
Monitor performance and adjust the slider based on your comfort level with AI-generated answers.
By fine-tuning this setting, you can ensure Fullview’s AI agent operates effectively within your trust parameters.