https://retellai.com/?utm_source=discord&utm_medium=seo logo
Powered by
Is there any way that an agent can backtrack to pr...
# ❓・ai-evy-support
w
@AI Evy Is there any way that an agent can backtrack to previous nodes if it needs to? the connectors can only be made one way which seems like a pretty big limitation
a
Thanks for tagging me @weronika. I'll look through relevant sources and get back to you shortly.
### Answer Yes, there are ways to handle backtracking in Conversation Flow agents, though it requires specific configuration. For Global Nodes: You can enable the "Go back to previous node" functionality. [(1) ]() When you set up a global node (which can be triggered from anywhere in your flow) , you need to configure a "Go Back Condition" that determines when the conversation should return to the original node. The key is properly defining this condition - for example, when a booking process is complete or a specific variable is set. However, this feature doesn't work automatically - you need to configure the exit condition that triggers the return. Without a properly defined "Go Back Condition," the component won't know when to send the conversation back to the main flow. Alternative Approach: According to community feedback, Retell doesn't automatically return from global nodes, but you can build this functionality by storing the last node in a variable and routing back manually with a Logic Split. [(2) ]() Workaround for Complex Flows: Some users have found that for scenarios requiring frequent backtracking between multiple nodes, using a single prompt agent with detailed instructions can be more reliable than conversation flow, especially when you need patterns like going from node 10 to node 3, or node 7 to node 2. [(3) ]() The limitation you're experiencing is real - edges in conversation flow are directional. [(4) ]() You'll need to either use global nodes with proper go-back conditions, manual routing with variables, or consider if a single prompt agent might
better suit your use case.
​ To learn more, see the sources I considered: [[1] "Go back to previous node" does not work on components or any other items that has no prompt in it.](https://discord.com/channels/1186424376300679209/1469335707335135252) [[2] fallback to the previous node from Global node](https://discord.com/channels/1186424376300679209/1388025209533235261) [[3] Discord message from @development_org_26839 on 2025-07-04](https://discord.com/channels/1186424376300679209/1186462578310910112/1390774938679378011) [[4] Conversation Flow Overview](https://docs.retellai.com/build/conversation-flow/overview) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
Thanks for the follow-up, @weronika. I'll get back to you soon.
w
@AI Evy how do i set up a existing node as a global node
a
### Answer To convert an existing node into a global node, you need to access the node settings. [(1) ]() [(2) ]() In the node configuration, there's a Global Node setting that you can enable. This setting is available in both Conversation Nodes and Logic Split Nodes. Once enabled as a global node, that node can be accessed from anywhere in your conversation flow based on the conditions you define for transitioning to it. Keep in mind that global nodes are designed to handle scenarios that can occur at any point in the conversation - like handling general questions, transfer requests, or other intents that might arise regardless of where the user is in the main flow. [(3) ]() [(4) ]()
​ To learn more, see the sources I considered: [[1] Logic Split Node](https://docs.retellai.com/build/conversation-flow/logic-split-node) [[2] Conversation Node](https://docs.retellai.com/build/conversation-flow/conversation-node) [[3] fallback to the previous node from Global node](https://discord.com/channels/1186424376300679209/1388025209533235261) [[4] Is it possible to return from the Global Node](https://discord.com/channels/1186424376300679209/1399484507530334368) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
Thanks for the follow-up, @weronika. I'll get back to you soon.
w
@AI Evy where do i acess the node settings?
a
### Answer To access node settings in Conversation Flow, click on the specific node you want to configure. [(1) ]() [(2) ]() For global settings that apply to the entire conversation (like voice, language, and LLM selection) , click on the empty canvas and then click the settings option. [(3) ]() Within individual node settings, you'll find options including: - Global Node toggle to enable/disable global node functionality [(1) ]() [(2) ]() - Block Interruptions to control whether the agent can be interrupted [(1) ]() - LLM selection to choose a different model for that specific node - Fine-tuning Examples for conversation response and transition tuning [(1) ]() [(2) ]() For Conversation Nodes specifically, you'll also see Skip Response settings, which when enabled, allows the agent to transition automatically after speaking. [(1) ]()
​ To learn more, see the sources I considered: [[1] Conversation Node](https://docs.retellai.com/build/conversation-flow/conversation-node) [[2] Logic Split Node](https://docs.retellai.com/build/conversation-flow/logic-split-node) [[3] Step 1: Configure global settings](https://docs.retellai.com/build/conversation-flow/global-setting) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
w
@AI Evy i dont have any of these settings Global Node toggle to enable/disable global node functionality (1) (2) Block Interruptions to control whether the agent can be interrupted (1) LLM selection to choose a different model for that specific node Fine-tuning Examples for conversation response and transition tuning in my prompt tree. all i can see is "Enable State Interruption Sensitivity"
a
Thanks for the follow-up, @weronika. I'll get back to you soon.
### Answer Based on the settings you're describing, it sounds like you might be using a Multi-Prompt agent rather than a Conversation Flow agent. [(1) ]() The settings you mentioned (Global Node toggle, Block Interruptions, LLM selection per node, and Fine-tuning Examples) are specific to Conversation Flow agents, which use a visual node-based editor. [(2) ]() [(3) ]() [(1) ]() Multi-Prompt agents have different settings, including "Enable State Interruption Sensitivity," which matches what you're seeing. To use Conversation Flow features: You'll need to create or switch to a Conversation Flow agent. Conversation Flow was introduced as a new agent type that provides more granular control over conversation paths with visual nodes and edges. [(3) ]() [(1) ]() If you need the backtracking functionality and global node features we discussed earlier, you'll need to work with a Conversation Flow agent rather than a Multi-Prompt agent.
​ To learn more, see the sources I considered: [[1] Step 1: Configure global settings](https://docs.retellai.com/build/conversation-flow/global-setting) [[2] Function Node Overview](https://docs.retellai.com/build/conversation-flow/function-node) [[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.
w
@AI Evy when i test a voice agent in a manual typed chat, is it normal for the agent to give overly long verbose responses with odd formatting? why does it do this and will it behave differently in a real call
a
Thanks for the follow-up, @weronika. I'll get back to you soon.
### Answer I don't have specific information in the available sources about differences between chat testing behavior and actual voice call behavior for Retell AI agents, or why agents might produce verbose or oddly formatted responses during manual typed chat testing. However, based on the available information, voice agents are optimized for natural spoken conversations with features like turn-taking models, voice activity detection, and speech synthesis that create human-like dialogue flow. [(1) ]() [(2) ]() These conversational dynamics differ significantly from text-based chat interactions. For specific guidance on chat testing behavior versus live call performance, I recommend: - Visit [Documentation]() - Visit [Hire certified Retell partner]() Would you like a general answer based on my knowledge instead?
​ To learn more, see the sources I considered: [[1] Speech Processing](https://www.retellai.com/glossary/speech-processing) [[2] Voice AI](https://www.retellai.com/glossary/voice-ai) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
PreviousNext
Powered by