I need some help,
I need to give a demo to the client to show how the voice calling agent works. I want to make it live for a short time with low latency. Could you please suggest how I can deploy it live using free resources?
t
tall-airline-16756
02/19/2026, 5:12 PM
Thanks for reaching out, Muhammad. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
tall-airline-16756
02/19/2026, 5:13 PM
Thanks — I understand you need to show a live, low-latency voice calling agent for a short client demo using free resources. Below is a clear, step-by-step plan you can follow quickly.
Quick summary of the approach
• Use LiveKit Cloud’s free tier for short demos.
• Build the agent in Agent Builder (no-code) and deploy instantly.
• Use LiveKit Inference (co-located model routing) for low latency STT/LLM/TTS.
• Optionally attach the free US phone number to demo inbound calls.
Step-by-step guide
1. Create a free LiveKit Cloud account
• Sign up for the free tier (no credit card). The free account includes session minutes suitable for short demos.
1. Build the voice agent with Agent Builder (fast, no-local-setup)
• Open Agent Builder in your browser and create the agent flow and voice settings.
• Use the built-in preview to test interaction in real time.
• Agent Builder uses the same production voice pipeline, so your demo will reflect live behavior.
• When ready, click Deploy agent to go live instantly.
1. Ensure low latency
• Deploy on LiveKit Cloud so the agent and inference services are co-located; this minimizes network hops and reduces latency.
• Use LiveKit Inference for STT, LLM, and TTS so speech and model calls run through provisioned, low-latency connections.
• Before the demo, run a few test calls and measure Time To First Token (TTFT) / Time To First Byte (TTFB) in the observability tools to confirm responsiveness.
1. (Optional) Enable phone calling for the demo
• You can use the one free US local phone number included in the plan.
• Attach a dispatch rule to route inbound calls to your deployed agent so the client can call a phone number rather than use the browser.
1. Test and prepare fallback options
• Test with colleagues from different networks to confirm latency and audio quality.
• Have a backup demo plan: a recorded session of the live agent or a short local demo using the preview mode in case network conditions degrade.
What I can help with next
• Walk you through Agent Builder: I can guide you step-by-step in the builder UI.
• Help configure a dispatch rule and attach the free phone number.
• Run a quick checklist to validate latency metrics before your demo.
Would you like me to walk you through Agent Builder now or help set up the phone number and dispatch rule for the demo?
Sources: LiveKit | Pricing | LiveKit | Brand | LiveKit | Announcing Agent Builder: a pathway to code | Agent Builder | LiveKit Documentation
a
ancient-telephone-52360
02/19/2026, 5:15 PM
Could you please share a setup video with me?
t
tall-airline-16756
02/19/2026, 5:16 PM
Thanks — I can share setup videos to help you deploy and demo the voice calling agent.
Here are two official walkthroughs that cover everything you’ll need:
• Add a Phone Number to Your AI Voice Agent —
https://youtu.be/KJ1CgZ0iZbY▾
Shows how to purchase a phone number in LiveKit, attach a dispatch rule, and route inbound calls to your agent (zero-to-ringing in ~60 seconds).
• Build an AI Voice Agent in Minutes (no code required) —
https://youtu.be/FerHhAVELto▾
Demonstrates using Agent Builder to create, preview, and deploy a production-grade voice agent from your browser.
Would you like a short checklist to follow while watching (so you can pause and complete steps), or would you prefer I walk you through the setup live via chat while you follow the video?
Sources: LiveKit DevDay | LiveKit | Introducing LiveKit Phone Numbers: zero to ringing in 60 seconds