Skip to main content
What is debugging? Debugging is the process of identifying, analysing, and fixing errors (known as bugs) in software programs or hardware systems, to make sure they work as intended. Why do we debug? After monitoring and auditing the call, it is time to solve the issues the agent faced on a live call — so the agent does not repeat the same mistake again and handles the customer better in future.
  1. Export the batch of calls. Export the batch of calls you want to debug to an Excel sheet. This sheet contains all the features and filters needed to find issues.
  2. Find the issues. After importing the batch you want to analyse, apply filters along various parameters (this is called funneling) — such as functions, post-call variables, dates, and time. Then go through each call and look for common patterns in issues, such as: the agent is unable to take the postal code because of a transcriber issue.
  3. Find the root cause. After spotting a common issue across multiple calls, find the source or root cause by going through the Call Log of each filtered call.

​
Call Log

A Call Log is a detailed record of everything that happened during a call — who spoke, what was said, any errors, and how the call ended. It is saved automatically by the system for review and debugging. To access Call Logs: go to Call History, apply filters, select the required call, then click Call Log (in the Call Details section). How to analyse call logs:
  • Copy the error statement (it could be a function, a variable, or a sentence).
  • From the call transcript, search for that exact statement in the call log (using Ctrl+F or Command+F).
  • Analyse the agent’s thinking around that statement and find the root cause of the error.
Tip: You can use logger.info or logger.warning while writing a function or variable. Think of logging like adding sticky notes throughout your code that say “I reached this point” or “something went wrong here.” When your function or variable runs, these notes get written to the call log — so when something breaks, you can read the notes and figure out exactly where and why it failed.
  1. Fix the issue. Once you know the root cause, fix it by updating the agent’s prompt, functions, or variables — whichever is causing the problem.

​
Investigate with AI FDE

Review Calls and Test Agents

Give AI FDE a call link or Call ID to investigate the transcript and available call logs in Ask mode.