> ## Documentation Index
> Fetch the complete documentation index at: https://docs.retellai.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Debug your agent response

Here's how to troubleshoot common issues:

## When the agent's response is unexpected

Step 1: Click the "Debug" button on the agent's response

<Frame>
  <img style={{height: "600px"}} src="https://mintcdn.com/retellai/pRGcctz_zOqy0mSt/images/test-llm/debug-response-utterance.png?fit=max&auto=format&n=pRGcctz_zOqy0mSt&q=85&s=c5563b86f24ef567900e89ec50732e34" alt="LLM Playground Debug" width="840" height="1804" data-path="images/test-llm/debug-response-utterance.png" />
</Frame>

Step 2: Follow the suggested solutions

<Frame>
  <img style={{height: "600px"}} src="https://mintcdn.com/retellai/pRGcctz_zOqy0mSt/images/response_popup.png?fit=max&auto=format&n=pRGcctz_zOqy0mSt&q=85&s=13863a792aeedadcc7080db2559dc14d" alt="LLM Playground Debug" width="483" height="513" data-path="images/response_popup.png" />
</Frame>

Available solutions:

1. Add fine-tuning examples ([guide](/build/conversation-flow/debug-guide#add-conversation-finetune-examples))
2. Split the node into multiple nodes ([guide](/build/conversation-flow/debug-guide#split-the-node-into-multiple-nodes))
3. Adjust the LLM temperature ([guide](/build/conversation-flow/debug-guide#adjust-the-llm-temperature))

Step 3: Click "Regenerate Answer" to get a new response

Step 4: Use "Regenerate 10 Times" to test response consistency

## When the agent doesn't transition correctly

Step 1: Click the "Debug" button on the agent's response

<Frame>
  <img style={{height: "600px"}} src="https://mintcdn.com/retellai/pRGcctz_zOqy0mSt/images/response_debug.png?fit=max&auto=format&n=pRGcctz_zOqy0mSt&q=85&s=916d5c1a5f7fd6367409f55fca84d835" alt="LLM Playground Debug" width="429" height="799" data-path="images/response_debug.png" />
</Frame>

Step 2: Click "Didn't transition as expected?"

Step 3: Follow the suggested solutions

<Frame>
  <img style={{width: "300px"}} src="https://mintcdn.com/retellai/pRGcctz_zOqy0mSt/images/transition_popup.png?fit=max&auto=format&n=pRGcctz_zOqy0mSt&q=85&s=2229ba49016f52bc024dcad615d3c31a" alt="LLM Playground Debug" width="475" height="305" data-path="images/transition_popup.png" />
</Frame>

Available solutions:

1. Add fine-tuning transition examples ([guide](/build/conversation-flow/debug-guide#add-transition-finetune-examples))
2. Split the node into multiple nodes ([guide](/build/conversation-flow/debug-guide#split-the-node-into-multiple-nodes))

Step 4: Click "Regenerate Transition" to get a new response

## When the agent transitions at the wrong time

Step 1: Click the "Debug" button on the transition dialog

<Frame>
  <img style={{height: "600px"}} src="https://mintcdn.com/retellai/pRGcctz_zOqy0mSt/images/test-llm/debug-transition-utterance.png?fit=max&auto=format&n=pRGcctz_zOqy0mSt&q=85&s=0d3d3251f44bfea4bacccb9363472b72" alt="LLM Playground Debug" width="786" height="1326" data-path="images/test-llm/debug-transition-utterance.png" />
</Frame>

Step 2: Follow the suggested solutions

<Frame>
  <img style={{width: "300px"}} src="https://mintcdn.com/retellai/pRGcctz_zOqy0mSt/images/transition_popup.png?fit=max&auto=format&n=pRGcctz_zOqy0mSt&q=85&s=2229ba49016f52bc024dcad615d3c31a" alt="LLM Playground Debug" width="475" height="305" data-path="images/transition_popup.png" />
</Frame>

Available solutions:

1. Add fine-tuning transition examples ([guide](/build/conversation-flow/debug-guide#add-transition-finetune-examples))
2. Split the node into multiple nodes ([guide](/build/conversation-flow/debug-guide#split-the-node-into-multiple-nodes))

Step 3: Click "Regenerate Transition" to get a new response

## When the agent's responses are inconsistent

Step 1: Click the "Debug" button on the agent's response

Step 2: Try these solutions:

1. Add fine-tuning examples ([guide](/build/conversation-flow/debug-guide#add-conversation-finetune-examples))
2. Split the node into multiple nodes ([guide](/build/conversation-flow/debug-guide#split-the-node-into-multiple-nodes))
3. Adjust the temperature ([guide](/build/conversation-flow/debug-guide#adjust-the-llm-temperature))

Step 3: Use "Regenerate 10 Times" to test response consistency

Step 4: Click "Regenerate 10 Times" to test response consistency

<Frame>
  <img style={{height: "600px"}} src="https://mintcdn.com/retellai/pRGcctz_zOqy0mSt/images/regen.png?fit=max&auto=format&n=pRGcctz_zOqy0mSt&q=85&s=bf0baeb92671cf637b5d7d01478bba32" alt="LLM Playground Debug" width="460" height="552" data-path="images/regen.png" />
</Frame>