### Answer When you duplicate an agent with shared components, those components may convert to local components, which can cause unexpected behavior if the original shared components still contain old conversation logic. [(1) ]() For conversation flow agents specifically, there's a known issue where changes may appear saved in the frontend but don't effectively apply during testing. [(2) ]() Even after deploying changes, the agent may still reference old configurations. Recommended fixes: 1. Verify your changes are actually deployed - Check that modifications show up when you export the flow as JSON, not just in the frontend. 2. Split complex nodes - If your agent has nodes with lengthy instructions that might contain remnants of old logic, break them into multiple simpler nodes for more consistent behavior. [(3) ]() 3. Review transition conditions - Ensure transition conditions precisely match your intended triggers and don't reference old conversation paths. 4. Add transition finetune examples - Provide examples to help the model understand your expectations and avoid falling back to old patterns. 5. Clear any lingering shared component references - If you duplicated an agent that used shared components, manually verify all components are truly local and updated. [(1) ]() If the issue persists after these steps, you may need to create a fresh agent rather than duplicating, as duplication can sometimes carry over hidden configuration remnants.

​ To learn more, see the sources I considered: [[1] Discord message from @momsramen on 2026-01-23](https://discord.com/channels/1186424376300679209/1186462578310910112/1464095837654094005) [[2] Issue with Saving and Deploying Changes in Conversation Flows](https://discord.com/channels/1186424376300679209/1361282824069714122) [[3] Debug guide](https://docs.retellai.com/build/conversation-flow/debug-guide) ​

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