### Answer The issue is likely related to how the Press Digit Node behaves—the agent will not speak when it's in this node. [(1) ]() A few things to check: 1. **Transition timing**: Transition happens when the user (in this case, the IVR system) is done speaking. Your transition condition should reflect when the digit press was successful and the IVR has confirmed acceptance—something like "Reached connected state" or "IVR confirmed call accepted." 2. **Detection delay**: You have 1500ms set, which should be reasonable. The documentation recommends setting this to around 1 second to ensure the whole IVR menu is captured before the agent makes a decision. 3. **Conversation node configuration**: Once you transition to your welcome/greeting node, make sure that node is configured to have the agent speak first. Check if your conversation flow has

start_speaker

set to

"agent"

. [(2) ]() 4. **Consider using "Skip response"**: On your Press Digit Node's outgoing transition to the welcome node, you might try enabling the "skip response" feature on the welcome conversation node, which can select a destination node to transition to when the agent is done speaking content of that node. [(3) ]() The core issue is that after transitioning out of the Press Digit Node, your welcome node may be waiting for user input before speaking. Ensure your welcome conversation node's instruction explicitly tells the agent to greet immediately upon entering.

​ To learn more, see the sources I considered: [[1] Press Digit Node](https://docs.retellai.com/build/conversation-flow/press-digit-node) [[2] Update Conversation Flow](https://docs.retellai.com/api-references/update-conversation-flow) [[3] 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.

### Answer The Skip Response setting is a node-level option on Conversation Nodes that changes how transitions work. [(1) ]() When Skip Response is enabled on a Conversation Node: - The transition will only have one edge that you can connect - When the agent is done talking, it will transition to the next node via that specific edge - This is useful when you want the agent to say things like disclaimers, where you don't need a response to move on to another node How this applies to your scenario: On your welcome/greeting Conversation Node (the one that comes after the Press Digit Node) , you could enable Skip Response and write a static greeting message. This would make the agent: 1. Immediately speak the greeting when entering the node 2. Automatically transition to the next node after speaking (without waiting for user input) This differs from the default behavior where the Conversation Node may wait for user input before transitioning. Alternative approach: If you don't want to use Skip Response, ensure your welcome node uses a Static Sentence instruction type rather than just a prompt—this makes the agent say a fixed sentence first upon entering the node.

​ To learn more, see the sources I considered: [[1] Conversation Node](https://docs.retellai.com/build/conversation-flow/conversation-node) ​

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

### Answer ### Answer The inconsistent behavior you're experiencing may be related to how Flex Mode interacts with your flow. There are some important considerations: Flex Mode Limitations: When Flex Mode is enabled, all nodes get converted into a single flex node that behaves like a single prompt agent. [(1) ]() The documentation notes some known issues: - LLM might not always follow the static text instruction - If you use flex mode for more than 20 nodes, performance might degrade and the agent might have higher hallucination risk - It's recommended to write node instructions in a concise manner so the LLM can better focus on the task Potential Issues: 1. **Flex Mode + Press Digit behavior**: Since Flex Mode converts everything to a single prompt, the Press Digit Node's specific timing and behavior may not work as reliably as in standard conversation flow mode. 2. **Detection delay**: Your 1500ms delay should be reasonable—the documentation recommends setting detection delay to around 1 second to ensure the whole IVR menu is captured. [(2) ]() However, if RingCentral's prompt is longer or has pauses, you may need to increase this. Recommendations: 1. Try disabling Flex Mode for this flow to see if standard conversation flow handles the Press Digit Node more reliably. 2. Increase detection delay beyond 1500ms if RingCentral's IVR prompt is slow or has pauses before the "press 1" instruction completes. 3. **Check your transition conditions**: The documentation states transition happens when the user is done speaking. You should write conditions like "Reached connected state" or handle scenarios like "Menu repeated xx times" for getting stuck in the IVR. The "I'm sorry I did not hear you" response fr

om RingCentral suggests the digit press either didn't happen or happened at the wrong time—this points to a timing/detection delay issue.

​ To learn more, see the sources I considered: [[1] Flex Mode](https://docs.retellai.com/build/conversation-flow/flex-mode) [[2] Press Digit Node](https://docs.retellai.com/build/conversation-flow/press-digit-node) ​

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

### Answer --- The issue is likely with your transition condition phrasing. For Press Digit Nodes, transition happens when the user (the IVR system) is done speaking, and the condition is evaluated based on what was heard. [(1) ]() Your current condition—"We have already pressed 1 to accept the call (digit press succeeded) , so proceed to greeting the caller"—reads more like an instruction than an observable condition. **Try rewriting the condition to describe what the agent should hear that indicates success:** -

"Call connected"

or

"Caller is now on the line"

-

"IVR confirmation heard"

or

"No more IVR prompts"

- Something that describes what RingCentral says after accepting the call Also consider adding a fallback condition for edge cases like getting stuck. The documentation suggests writing conditions like

"Menu repeated xx times"

to handle scenarios where the IVR loops. Debugging tip: If the agent isn't transitioning even though the digit press works, review the call transcript to see what the agent "heard" after pressing 1. Your transition condition needs to match what RingCentral actually says upon successful connection. [(2) ]() If you continue seeing issues, you might also try adding a transition finetune example to help the model understand when to transition.

​ To learn more, see the sources I considered: [[1] Press Digit Node](https://docs.retellai.com/build/conversation-flow/press-digit-node) [[2] 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 --- Since the issue is silence after the digit press (no IVR audio to trigger a transition) , the transition may never fire because the condition is evaluated when the user is done speaking—but if there's silence, there's nothing to evaluate against. [(1) ]() For a finetune transition example, you would provide: [(2) ]() 1. A transcript as context showing what happens during the call (including the IVR prompt and the silence after pressing 1) 2. The transition result you want (i.e., which node to transition to) You can select

user

,

agent

, or

function

as the role for each line of the transcript. Example finetune setup: - **Transcript context**: Include the IVR saying "To accept the call, press one" followed by silence (or whatever RingCentral says) - **Transition result**: Select your welcome/greeting node as the destination Additional consideration: The core problem may be that the Press Digit Node waits for the "user" (IVR) to finish speaking before evaluating transitions. [(1) ]() If RingCentral goes silent after you press 1, the node may be stuck waiting. You might also try adding a condition like

"Silence detected after digit press"

or

"No response from IVR"

to handle the silence scenario explicitly. If you cannot distinguish between the different nodes available as transition targets when setting up finetune examples, try renaming your nodes to make them easier to distinguish. [(2) ]()

​ To learn more, see the sources I considered: [[1] Press Digit Node](https://docs.retellai.com/build/conversation-flow/press-digit-node) [[2] Finetune Examples](https://docs.retellai.com/build/conversation-flow/finetune-examples) ​

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