https://retellai.com/?utm_source=discord&utm_medium=seo logo
Powered by
is there a free voice that I could test using 'Tes...
# ❓・ai-evy-support
a
Thanks for tagging me @Q4. I'll look through relevant sources and get back to you shortly.
### Answer I don't have specific information in the available sources about which voices are free for "Test Audio" testing. What I can tell you is that Web Call/Phone Call Testing (which includes Test Audio) is listed as having "Call charges" as its cost, while LLM Playground and LLM Simulation testing are listed as "Free." [(1) ]() For details on free voice options or trial minutes, I'd suggest: - Check out these potentially useful sources: [Testing Overview]() - Visit [Contact Sales]() for specific pricing questions Would you like me to try to provide a general answer based on my knowledge instead?
​ To learn more, see the sources I considered: [[1] Testing Overview](https://docs.retellai.com/test/test-overview) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
q
@AI Evy can conversasion return back to previuos node when it reached global node
a
Thanks for the follow-up, @Q4. I'll get back to you soon.
### Answer No, Retell AI currently does not support returning from a Global Node to the original conversation flow position. [(1) ]() This is described as an architectural limitation. When a conversation transitions to a Global Node, it cannot automatically return to the exact node it came from. As a workaround, you can transfer from a regular node to a Global Node, but not vice versa in a way that resumes the previous position. If this feature is important for your use case, you can submit a feature request in the #📒・feature-requests channel.
​ To learn more, see the sources I considered: [[1] 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, @Q4. I'll get back to you soon.
q
@AI Evy is there a universal prompt for email capture?
a
### Answer ### Answer There isn't an official "universal" prompt from Retell AI documentation, but a community member shared a detailed email collection prompt that has worked well for them. [(1) ]() Here's the prompt approach:
Copy code
Email addresses will always be in the format of Name@domain.com. Examples include: (karan@yahoo.com; lisa@response.ai; michael.masters@info.net) . When confirming an email address, respond in this format -"Just to make sure I've got it right, your email address is Karan at Yahoo dot com [Make sure to spell out each character of the email individually, unless it is a Gmail or Yahoo domain]." 
    If the caller says you got the email wrong, apologize for the mistake and ask the user to spell it out. Respond by spelling it back to them. 
    For example, the caller says "K A R A N @ Y A H O O dot com." You will respond with, "Just to make sure I've got it right, your email address is K A R A N at yahoo dot com."
    For example, the caller says "K A R A N @ G M A I L dot com." You will respond with, "Just to make sure I've got it right, your email address is K A R A N at Gmail dot com."
    Another example: the caller says "Karan dot Sharma at Gmail dot com." You will respond with, "K A R A N dot S H A R M A at Gmail dot com."
    Another example: the caller says "Karan underscore Sharma at Yahoo dot com." You will respond with, "K A R A N underscore S H A R M A at Y A H O O dot com."

Words like "At," "Dot," "Dash," "Gmail" and "Underscore" should not be spelt out. For example, if someone says "Karan.Sharma," you will spell it out as "K A R A N dot S H A R M A."
    Wait for the caller to respond. Repeat the email address properly. If the user says the email address you mentioned is wrong, ask for clarification and prompt the use for details such as "can you please spell it out for me?" For letters that are difficult to transcribe, you can ask clarifying questions such as "is that P as in Paul" or "B as in boy" or "Z as in zebra" or "C as in carrot". Only proceed when the user confirms the correct email address.
If using Conversation Flow builder, add this to the prompt first: 
Collect and verbally confirm the correct spelling of the users email address;
and for the transition use: 
Only proceed when the user confirms the correct email address.
This approach handles common issues like P/B confusion and custom domains by prompting the agent to spell out characters and ask clarifying questions.
​ To learn more, see the sources I considered: [[1] Email collection prompt](https://discord.com/channels/1186424376300679209/1349712876243390526) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
Thanks for the follow-up, @Q4. I'll get back to you soon.
q
@AI Evy is there a known prompt how to collect customer's name
a
### Answer I don't have a specific "universal" name collection prompt in the available sources like the email prompt shared earlier. However, based on the Prompt Engineering Guide, here's a recommended approach using sectional prompts: [(1) ]()
Copy code
## Task Instructions
1. Greet the caller warmly
2. Ask for their full name
3. Confirm understanding by paraphrasing back: "Just to confirm, your name is [Name], is that correct?"
4. For names that are difficult to transcribe, ask clarifying questions such as "Could you spell that for me?" or "Is that S as in Sam?"
5. Only proceed when the caller confirms the correct name
Key best practices from the documentation: - Ask one question at a time – avoid overwhelming the caller with multiple questions - Confirm understanding – paraphrase important information back to the caller - Be conversational – use natural language and acknowledge what the caller says If you're using Conversation Flow, you could structure a dedicated state for name collection (similar to the "Lead Qualification" example where State 1 gathers customer information before proceeding) . [(2) ]() Would you like me to try providing a more detailed general answer based on my knowledge?
​ To learn more, see the sources I considered: [[1] Prompt Engineering Guide](https://docs.retellai.com/build/prompt-engineering-guide) [[2] Single/Multi Prompt Agent Overview](https://docs.retellai.com/build/single-multi-prompt/prompt-overview) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
q
@AI Evy Flex mode in Conversasional pathways. What are Pros and cons?
a
Thanks for the follow-up, @Q4. I'll get back to you soon.
### Answer ### Answer Flex Mode combines conversation flow structure with single prompt flexibility for dynamic task handling. [(1) ]() Pros: - You can easily switch context between different tasks (every node becomes like a global node) - The agent can move to the proper task if the user completes multiple tasks at the same time - After switching context to another flow, the agent can resume the previous task without repeating already completed steps - Gives you the clarity of a flowchart (business steps) with the freedom of a single prompt Cons / Known Issues: - Node-level Knowledge Base is ignored in Flex Mode — you must configure KB at the agent level - No "Speak After Execution" setting — the agent always speaks after function execution - No "waitForResult" setting — the agent always waits for the function to complete - LLM may not always follow static text instructions - Avoid using Equation edges — LLMs are bad at interpreting equation conditions and may produce weird behaviors - Performance may degrade with more than 20 nodes, and hallucination risk increases — splitting into smaller components is recommended - A "race condition" can cause the end call tool to terminate before TTS finishes speaking (workaround: move goodbye logic to Global Prompt instead of using "Speak during execution") [(2) ]() Best Practices: - Write node instructions concisely so the LLM can better focus on the task [(1) ]() - Use only Prompt edges - Be explicit on transitions with crisp, observable conditions
​ To learn more, see the sources I considered: [[1] Flex Mode](https://docs.retellai.com/build/conversation-flow/flex-mode) [[2] Speak During Execution not working for End Call node](https://discord.com/channels/1186424376300679209/1451006517909258352) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
PreviousNext
Powered by