Kore.ai provides the ability for a human reviewer (developer, customer, support personnel, etc.) to passively review every bot message and choose to train the bot. To get a quick overview of how a bot is performing, developers can review the Chat Logs in the Kore.ai Bot Builder. In the Chat Logs, developers see a summary of successful and failed scenarios.
- Success Scenarios – The user input for the bot successfully matched a task.
- Failed Scenarios – The user input for the bot did not match a task. For a failed scenario, developers should determine why the interaction with the bot failed and optimize bot settings to increase the success rate to enhance customer satisfaction.
A developer can review all failed scenarios and determine if the bot needs to have recognition performance enhancements and train using ML. Not all failed scenarios need to be fixed. Some user input is too ambiguous and requires more dialog asking for additional information.