### Answer --- An agent repeating itself instead of transitioning typically stems from a few common issues: ## Unclear Transition Conditions The LLM may not understand when to move to the next step. [(1) ]() Your prompt needs explicit triggers—without clear conditions, the agent gets stuck repeating the same step. [(2) ]() ## Too Many Transition Options Having 4-5+ transitions from a single state can confuse the LLM, making it difficult to decide which path to take. Consider consolidating similar transitions or splitting decisions across multiple steps. ## Vague Step Boundaries If your prompt doesn't clearly define when one step ends and the next begins, the agent won't know to proceed. Break prompts into smaller, numbered steps with explicit progression logic. ## Recommended Fixes 1. Add explicit decision points in your prompt:

1. Greet the client and ask how you can help.
2. After the caller's response, classify their intent into:
   - Personal injury → transition to personal_injury
   - Workplace injury → transition to workplace_injury
   (Do not say this step out loud.)

2. Use trigger words — list specific words/phrases that should trigger transitions. [(1) ]() 3. For Conversation Flow agents, consider splitting complex nodes into multiple simpler nodes for more consistent behavior. [(3) ]() 4. Add transition finetune examples to help the model understand your expectations. 5. Try a more capable LLM model if instructions are concise but the agent still struggles.

​ To learn more, see the sources I considered: [[1] Prompt Engineering Guide](https://docs.retellai.com/build/prompt-engineering-guide) [[2] Not transitioning between states](https://discord.com/channels/1186424376300679209/1291114413662801960) [[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.

### Answer For a node where the agent is supposed to take a note (e.g., collect information) , you'd use a prompt-based transition condition that checks whether the user has provided the required information. [(1) ]() ## Example Transition Conditions Prompt conditions work well for this scenario: -

User responds to question of their age

-

User said something about booking a meeting

-

User claims to be over 18

For a note-taking node specifically, you might write something like: -

User has provided the information requested

-

User responds with the details asked for

## Using Equation Conditions If you're checking whether a dynamic variable has been captured, you can use an **equation condition**:

{{variable_name}} exists

This returns true if the variable is defined and has a value—useful for confirming that information was successfully extracted before transitioning. ## Best Practice To ensure smooth transitions, cover all possible cases in your transition conditions. For a note-taking node, you'd want conditions for: 1. When the user provides the information → transition to next step 2. When the user declines or says something unrelated → handle appropriately (possibly via a global node)

​ To learn more, see the sources I considered: [[1] Step 3: Add transition conditions](https://docs.retellai.com/build/conversation-flow/transition-condition) ​

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