https://retellai.com/?utm_source=discord&utm_medium=seo logo
how do most people handle outbound calls where you...
# ❓・ai-evy-support
h
@AI Evy how do most people handle outbound calls where you need to disclose that you are recording
a
Thanks for tagging me @harounans. I'll look through relevant sources and get back to you shortly.
### Answer For outbound calls requiring recording disclosure, the common approach is to have your AI agent introduce itself and request consent at the start of the call. [(1) ]() Here are some example introductions you can use in your agent's prompt: - Example 1: "Hello, I'm Anna, your AI assistant on this recorded line. May I know who I'm speaking with today?" - Example 2: "Hi there! This is Sam, an automated assistant. Please be aware that this call is being recorded for quality assurance. Can I have your name, please?" - Example 3: "Good day! I'm Alex, an AI assistant on this call. For compliance purposes, could you please state your name and confirm that you are aware this call is being recorded?" Key considerations: - In one-party consent states (like New York, Georgia, Alabama) , only one party needs to be aware of the recording. - In two-party consent states (like California, Florida, Illinois, Pennsylvania) , all parties must give explicit consent before recording. If a caller does not consent, you should handle it gracefully—either end the call politely or offer alternative communication methods that don't involve recording.
​ To learn more, see the sources I considered: [[1] How to Legally Record Phone Calls in the US | Retell AI](https://www.retellai.com/blog/legally-recording-phone-calls-in-the-us-how-does-it-work) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
Thanks for the follow-up, @harounans. I'll get back to you soon.
h
What are some best practices for outbound calls @AI Evy
a
### Answer Here are some best practices for outbound calls with Retell AI: Setup & Configuration - Bind your agent to a phone number before making calls—configure both inbound and outbound agents as needed [(1) ]() - Use different agents for inbound vs outbound if your use cases require specialized handling - Ensure phone numbers are in E.164 format (e.g.,
+14157774444
) CPS (Calls Per Second) Management - Start with low CPS limits and scale gradually based on actual needs - Default CPS is 1 for most providers; Twilio max is 5, Telnyx max is 16 - Implement retry logic with exponential backoff to handle throttling gracefully
Copy code
javascript
const maxRetries = 3;
let retryDelay = 1000; // Start with 1 second
for (let i = 0; i < maxRetries; i++) {
  try {
    await makeCall() ;
    break;
  } catch (error) {
    if (error.code === 'rate_limited') {
      await sleep(retryDelay) ;
      retryDelay *= 2; // Exponential backoff
    }
  }
}
Personalization - Use
retell_llm_dynamic_variables
to inject customer-specific data (e.g., name, account info) into prompts , [(2) ]() - Store custom metadata with calls for tracking purposes Monitoring - Set up webhooks for real-time notifications on call events (
call.started
,
call.ended
,
call.analyzed
) [(1) ]() - Use the Get Call API to retrieve transcripts, recordings, and latency metrics Number Reputation - Warm up your numbers before high-volume campaigns to avoid being flagged as spam [(3) ]() - Consider using [Branded Caller ID]() to display your business name and increase answer rates [(5) ]()
​ To learn more, see the sources I considered: [[1] Outbound Calls (Make Calls)](https://docs.retellai.com/deploy/outbound-call) [[2] Create Phone Call](https://docs.retellai.com/api-references/create-phone-call) [[3] Discord message from @qcumbers1_cc on 2025-01-09](https://discord.com/channels/1186424376300679209/1294891368505937952/1326872454764695602) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
h
@AI Evy what about from a conversational experience perspective? Especially around welcome greetings for outbound campaigns
a
Thanks for the follow-up, @harounans. I'll get back to you soon.
### Answer ### Conversational Design for Outbound Welcome Greetings Key Elements for Effective Greetings [(1) ]() - Clear Entry Points — Define how conversations begin, setting tone and expectations immediately - Intent Clarity — Help users articulate needs naturally, without feeling constrained - Guided Prompts — Offer helpful options without sounding overly scripted or robotic Personalization with Dynamic Variables [(2) ]() Use dynamic variables to craft highly personalized greetings. Instead of a generic greeting, the AI can say something like: "Welcome back, [Customer Name]! We see you have an appointment scheduled for [Date] at [Time]." Structuring Your Outbound Flow [(3) ](

▾

) For outbound campaigns, consider using multi-state prompts where: - Your global state prompts set the identity, core principles, and global rules for the agent - Each state is specific to one action or one tool call - The welcome state handles the initial greeting, then transitions based on whether it's an existing client or new prospect Handling Transition Timing [(4) ]() If you need a brief pause after the greeting (e.g., to allow the caller to request a language switch) , you can: - Set the "Reminder Message Frequency" to 1-2 seconds - Use a single prompt agent for the first part of the greeting, then transfer with a delay before continuing Graceful Error Recovery [(1) ]() Handle misunderstandings smoothly without frustrating the caller—this is especially important for outbound where you're interrupting someone's d
ay.
​ To learn more, see the sources I considered: [[1] Conversational Design](https://www.retellai.com/glossary/conversational-design) [[2] AI for After-Hours Call Answering Services | Retell AI](https://www.retellai.com/blog/transforming-after-hours-call-answering-services-with-ai-phone-agents) [[3] Speed to Lead: Driving Results with Outbound Facebook Campaigns](

https://www.youtube.com/watch?v=9fgelfy6kQA▾

) [[4] Transition Between Welcome message and First Task is Greetings is taking more than 15 seconds](https://discord.com/channels/1186424376300679209/1401689973677363201) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
h
@AI Evy how should we handle voicemail
a
Thanks for the follow-up, @harounans. I'll get back to you soon.
### Answer ### Handling Voicemail in Outbound Campaigns Enable Voicemail Detection [(1) ]() Navigate to your agent settings → call settings and enable "Voicemail Detection". Choose Your Voicemail Behavior Select one of two options: - **Hang up**: The agent will disconnect when it detects a voicemail - **Leave a message**: The agent will leave a message after it's the agent's turn to speak. Dynamic variables are supported here for both static message and prompt: - **Static message**: Pre-recorded message like
Hi, this is {{your name}}, please give us a callback tomorrow at 10am.
- **Prompt**: Dynamically generated message based on instructions like
summarize the call and ask user to call back
Important Considerations - Voicemail detection only runs for the first 3 minutes of the call - If the call is detected as voicemail and the action is taken successfully, the call will have a disconnection reason of
voicemail_reached
- Latency impact is generally under 100ms - If the voicemail keeps talking in a loop, the agent will still hang up when detected - However, if you've selected "leave a message," the agent will only leave a message when it's the agent's turn to speak Timeout Configuration [(2) ]() If voicemail detection isn't working as expected, try increasing the detection duration—if duration is set to 10s, it won't detect after 10s.
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
​ To learn more, see the sources I considered: [[1] Handle voicemail](https://docs.retellai.com/build/handle-voicemail) [[2] Voicemail Detection is not working.](https://discord.com/channels/1186424376300679209/1319228307455348768) ​
h
@AI Evy what’s the common prompt engineering lifecycle for testing outbound agents to make sure we test against different cases like voicemail, hitting an actual customer, etc
a
Thanks for the follow-up, @harounans. I'll get back to you soon.
​ To learn more, see the sources I considered: [[1] Prompt Engineering](https://www.retellai.com/glossary/prompt-engineering) ​
### Answer ### Answer I don't have detailed documentation specifically covering the prompt engineering lifecycle or simulation testing workflows for outbound agents in the available sources. However, here's what I can share from the sources: Prompt Engineering Fundamentals [(1) ]() Good prompt engineering ensures AI voice agents: - Respond professionally and predictably across thousands of conversations - Understand what to prioritize in multi-intent or ambiguous situations - Stay within compliance and brand tone requirements - Guide users toward business outcomes Key areas to address in your prompts: - Tone and Personality Framing — Define whether the agent sounds formal, casual, empathetic, etc. - Behavioral Boundaries — Limit what topics or actions the AI can engage with - Error Handling Guidance — Prepare the AI to apologize gracefully, escalate when needed, or reframe questions if misunderstandings occur For Testing Different Scenarios Retell AI offers simulation testing capabilities that allow you to run batch or single-scenario tests with synthetic conversations to surface edge cases pre-deployment. However, detailed documentation on configuring these tests for specific cases (voicemail vs. live customer) isn't covered in my current sources. Check out these potentially useful resources: - [Prompt Engineering]() - [Create Voice Agent]() - [Update Voice Agent]() Would you like me to try providing a general answer based on common practices for testing voice AI agents?
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
h
@AI Evy are there any outbound templates you can recommend. I think the agents are inbound templates
a
Thanks for the follow-up, @harounans. I'll get back to you soon.
### Answer --- I don't have specific outbound agent templates in my available sources. The documentation covers the technical setup for outbound calls—binding agents to phone numbers, call parameters, and API configuration—but doesn't include pre-built prompt templates specifically designed for outbound use cases. [(1) ]() What I can confirm for outbound setup: - You can assign different agents for inbound vs outbound on the same number - Use
override_agent_id
to swap agents per-call without changing your default configuration - Inject personalization via
retell_llm_dynamic_variables
(e.g.,
{ "name": "John" }
) Check out these potentially useful resources: - [Outbound Calls (Make Calls) ]() - [Create Phone Call API]() Would you like me to try providing a general answer on common outbound prompt structures based on my knowledge?
​ To learn more, see the sources I considered: [[1] Outbound Calls (Make Calls)](https://docs.retellai.com/deploy/outbound-call) ​
Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.