### Answer Yes, you can use one phone number with multiple agents for outbound calls. There are two approaches: 1. Weighted Random Selection You can configure

outbound_agents

with weights when creating or updating a phone number. If set and non-empty, one agent will be picked randomly for each outbound call, with probability proportional to the weight. Total weights must add up to 1. [(1) ]() 2. Override Agent ID Per Call If you need to specify a particular agent for each outbound call (rather than random selection) , you can override the agent ID when initiating the call. The

outbound_agent_id

on the phone number serves as a default, but you can override it per call. Here's an example of the

outbound_agents

configuration:

json
"outbound_agents": [
  {
    "agent_id": "oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD",
    "weight": 0.5,
    "agent_version": 1
  }
]

[(2) ]() If not set or empty, the system falls back to

outbound_agent_id

. [(1) ]()

​ To learn more, see the sources I considered: [[1] Create Phone Number](https://docs.retellai.com/api-references/create-phone-number) [[2] Get Phone Number](https://docs.retellai.com/api-references/get-phone-number) ​

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

### Answer The

outbound_agents

field goes in the request body when you create or update a phone number—not in the Create Phone Call request. [(1) ]() [(2) ]() To set it when creating a phone number:

POST /create-phone-number

[(1) ]() To update an existing phone number:

PATCH /update-phone-number/{phone_number}

[(2) ]() Include

outbound_agents

in the JSON body like this:

json
{
  "outbound_agents": [
    {
      "agent_id": "oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD",
      "weight": 0.5,
      "agent_version": 1
    }
  ]
}

[(1) ]() If you need to specify a particular agent for a single outbound call instead, use

override_agent_id

in the Create Phone Call request body. [(3) ]()

​ To learn more, see the sources I considered: [[1] Create Phone Number](https://docs.retellai.com/api-references/create-phone-number) [[2] Update Phone Number](https://docs.retellai.com/api-references/update-phone-number) [[3] Create Phone Call](https://docs.retellai.com/api-references/create-phone-call) ​

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

Thanks for the follow-up, @Claude. I'll get back to you soon.

### Answer The

outbound_agents

with weights cannot be added to the Create Phone Call request body. Your JSON is for creating a phone call, but the weighted agents configuration belongs on the phone number itself, not on individual calls. [(1) ]() [(2) ]() You have two options: *Option 1: Configure weights on the phone number (separate request) * Update your phone number with the weighted agents:

json
{
  "outbound_agents": [
    {
      "agent_id": "oBeDLoLOeuAbiuaMFXRtDOLriTJ5tSxD",
      "weight": 0.5,
      "agent_version": 1
    }
  ]
}

[(2) ]() Then your Create Phone Call request stays as-is (the system will randomly select from your configured agents) . Option 2: Use

override_agent_id

per call If you need to specify a particular agent for each individual call, keep using

agent_id

in your Create Phone Call request (which acts as the override) . But this doesn't support weights—it's a direct agent selection. [(3) ]() The weights feature is designed for random distribution across multiple agents at the phone number level, not for per-call configuration. [(1) ]()

​ To learn more, see the sources I considered: [[1] Import Phone Number](https://docs.retellai.com/api-references/import-phone-number) [[2] Get Phone Number](https://docs.retellai.com/api-references/get-phone-number) [[3] Outbound Phone Call Not Connecting](https://community.retellai.com/t/outbound-phone-call-not-connecting/249) ​

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

### Answer No, you don't need to buy different numbers for each agent. You can use one phone number and specify which agent to use on a per-call basis using

override_agent_id

. [(1) ]() When making an outbound call, include the

override_agent_id

parameter to select the specific agent for that call:

javascript
const registerCallResponse = await retell.call.createPhoneCall({
  from_number: '+14157774444', // replace with the number you purchased
  to_number: '+12137774445', // replace with the number you want to call
  // Optional: per-call agent selection and overrides
  override_agent_id: 'agent_abc123',
  override_agent_version: 0, // or omit to use latest
}) ;

This way, for each call you make, you can dynamically choose: - Your restaurant agent for restaurant-related calls - Your barber agent for barber-related calls The

override_agent_id

parameter lets you override the agent used for that specific call. So your workflow would be: 1. Buy one phone number 2. When initiating each outbound call, pass the appropriate

override_agent_id

based on the business type (restaurant vs barber)

​ 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.

### Answer Here are key ways to reduce latency in your Retell AI voice agent: ## Model and TTS Optimization **Use a faster LLM**: Switch to a more efficient language model to reduce response time. [(1) ]() **Use a faster TTS**: Implement a faster text-to-speech system for quicker audio output and smoother conversations. ## Turn-Taking Optimization Implement a more sophisticated turn-taking model to accurately detect the end of a user's turn. This prevents the AI from interrupting or prematurely responding, leading to more natural conversations. Retell AI's turn-taking model uses predictive timing to ensure smooth transitions between speakers, combining sound signals with semantic understanding. [(2) ]() ## Knowledge Base Considerations Large knowledge bases embedded directly in prompts can impact performance. If you have large files in your base prompt, consider alternative approaches. [(3) ]() ## Performance Targets Retell AI achieves sub-600ms latency performance with optimized configurations. [(4) ]() [(5) ]() Key benchmarks include: - Time-to-First-Token: 180ms average [(5) ]() - End-to-End Latency: 620ms average - Barge-in Response: 140ms average ## Additional Notes Phone calls typically have slightly higher latency than web calls—around 200ms more is expected. [(6) ]() Non-standar

d conversation patterns (like rapid single-word responses) can cause the turn-taking model to wait longer, increasing perceived end-to-end latency.

​ To learn more, see the sources I considered: [[1] Troubleshooting AI Voice Agent Issues | Retell AI](https://www.retellai.com/blog/troubleshooting-common-issues-in-voice-agent-development) [[2] VAD vs. Turn-Taking Endpoints in Conversational AI | Retell AI](https://www.retellai.com/blog/vad-vs-turn-taking-end-point-in-conversational-ai) [[3] Performance-related configurations and noticeable latency during live calls](https://community.retellai.com/t/performance-related-configurations-and-noticeable-latency-during-live-calls/747) [[4] ](https://www.retellai.com/resources/retell-ai-receptionist-understands-calls-better) [[5] ](https://www.retellai.com/resources/ai-voice-agent-latency-face-off-2025) [[6] High end to end latency in phone calls](https://discord.com/channels/1186424376300679209/1450956304288845855) ​

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

### Answer --- I don't see a screenshot or specific latency value in your message. Could you share your current latency numbers? For reference, here are the benchmarks to compare against: Acceptable latency targets: - Production voice AI agents typically aim for 800ms or lower latency to maintain conversational flow. [(1) ]() - Humans expect near-instantaneous responses, usually within 300-500 milliseconds. [(2) ]() Retell AI's typical performance: - End-to-End Latency: 620ms average [(1) ]() - Time-to-First-Token: 180ms average Key metrics explained: - **P90 (90th Percentile) **: 90% of calls have latency below this value. [(3) ]() - **Median (50th Percentile) **: Half of the calls have latency less than this value. If your latency is consistently above 800ms, it may feel unnatural to callers. Latency exceeding 1000ms+ often leads to conversation overlap, reduced trust, and abandoned interactions. [(1) ]() Share your latency metrics and I can help assess whether optimization is needed.

​ To learn more, see the sources I considered: [[1] ](https://www.retellai.com/resources/ai-voice-agent-latency-face-off-2025) [[2] Issues with Portuguese (PT-PT) Voice Agent: Latency, Audio Recognition, and Email Capture](https://community.retellai.com/t/issues-with-portuguese-pt-pt-voice-agent-latency-audio-recognition-and-email-capture/711) [[3] Check actual latency](https://docs.retellai.com/reliability/check-actual-latency) ​

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

### Answer A latency of 1045-1425ms is higher than ideal and will likely feel unnatural to callers. [(1) ]() Why this matters: - Production voice AI agents typically aim for 800ms or lower latency to maintain conversational flow. - Humans expect near-instantaneous responses, usually within 300-500 milliseconds. [(2) ]() - Beyond 800ms, users begin to notice delays, leading to conversation overlap, reduced trust, and abandoned interactions. [(1) ]() Your latency won't necessarily sound "robotic," but it will feel **slow and stilted**—callers may assume the system didn't hear them and start speaking again, causing overlap. Troubleshooting steps: 1. Check for features marked with 🐢 in your configuration and disable some to reduce latency. [(3) ]() 2. **Monitor your LLM response time**—normal range is 500ms-900ms. If consistently above 900ms, use the fast tier or switch LLM providers. 3. **Try a different TTS provider**—one community member with similar latency issues was advised to switch from Cartesia Sonic-2 to Sonic-3 or ElevenLabs. [(4) ]() 4. **Consider geographic distance**—international calls add latency. Use phone numbers closer to your target audience. [(3) ]() 5. **Check if turn-taking is contributing**—brief responses can cause the system to wait longer before responding, adding to perceived latency. [(5) ](

8845855)

​ To learn more, see the sources I considered: [[1] ](https://www.retellai.com/resources/ai-voice-agent-latency-face-off-2025) [[2] Issues with Portuguese (PT-PT) Voice Agent: Latency, Audio Recognition, and Email Capture](https://community.retellai.com/t/issues-with-portuguese-pt-pt-voice-agent-latency-audio-recognition-and-email-capture/711) [[3] Troubleshoot high latency](https://docs.retellai.com/reliability/troubleshoot-latency) [[4] Slow Latency (3000ms, triple what it should be)](https://community.retellai.com/t/slow-latency-3000ms-triple-what-it-should-be/591) [[5] High end to end latency in phone calls](https://discord.com/channels/1186424376300679209/1450956304288845855) ​

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