https://retellai.com/?utm_source=discord&utm_medium=seo logo
I have a subagent that keeps transitioning to the ...
# ❓・ai-evy-support
s
@AI Evy I have a subagent that keeps transitioning to the begin node
a
Thanks for tagging me @Sanjay. I'll look through relevant sources and get back to you shortly.
### Answer When a subagent node keeps transitioning back to the begin node, this is often caused by issues with transition conditions or the node configuration. A similar issue was reported in the community where a conversation flow agent would loop back to the beginning node after every response. [(1) ]() Here are some troubleshooting steps: Check your transition conditions: [(2) ]() - Ensure your edges from the subagent node have proper transition conditions defined - Verify that the conditions can actually be met based on the conversation context - Remember that transitions can happen when the user is done speaking or when Skip Response is enabled and the agent finishes speaking Review your node instruction: - Make sure your subagent node instruction clearly defines when the task is complete - The instruction should guide the LLM on when to use tools and when the conversation in this node should end Verify tool configuration: - If you have tools attached, ensure they're properly configured with execution speech settings - Check if "Speak After Execution" is set appropriately for your use case Test in an actual call: [(1) ]() - One community member found that the issue appeared during manual text testing but worked fine in actual calls - Try testing with a real phone call to see if the behavior differs If the issue persists, I recommend sharing your agent ID with the Retell support team for further investigation.
​ To learn more, see the sources I considered: [[1] Getting stuck in a loop when using Agent Transfer with Conversional Flow.](https://discord.com/channels/1186424376300679209/1476544511156031538) [[2] Subagent Node](https://docs.retellai.com/build/conversation-flow/subagent-node) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.