### Answer Fine-tuning examples for the Extract Dynamic Variable node help train the model on when and how to transition out of that node. [(1) ]() Specifically, fine-tuning examples can be used to show the agent expected behaviors—for instance, when a user provides information upfront (like their name at the start of a call) , the agent should still correctly transition to the Extract Variable node rather than getting stuck. [(2) ]() This is useful for handling edge cases where users provide information in unexpected ways or at different points in the conversation, ensuring the extraction and transition logic works as intended.

​ To learn more, see the sources I considered: [[1] Extract Dynamic Variable Node](https://docs.retellai.com/build/conversation-flow/extract-dv-node) [[2] Stucks in conversation node](https://discord.com/channels/1186424376300679209/1462438307588739103) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer Correct—fine-tuning examples for the Extract Dynamic Variable node are specifically for tuning transitions, not for improving the actual extraction accuracy. [(1) ]() The node settings show that fine-tuning examples "can finetune transition." To improve extraction accuracy itself, focus on: - **Variable Description**: Write a clear, specific description of what the value should be - **Variable Type**: Choose the appropriate type (Text, Number, Enum, or Boolean) - **LLM Selection**: You can choose a different model for the node, which affects function argument generation If extractions are failing, common causes include overly long user responses causing the model to lose track, or unclear variable descriptions. [(2) ]()

​ To learn more, see the sources I considered: [[1] Extract Dynamic Variable Node](https://docs.retellai.com/build/conversation-flow/extract-dv-node) [[2] Discord message from @mauvatar on 2026-03-03](https://discord.com/channels/1186424376300679209/1186462578310910112/1478494393018355845) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.