Voice Agents Handbook

Skip the wasted weekend.

Building production voice AI with LiveKit

by Mahimai Raja J

Nine chapters. Three surfaces: phone, browser, mobile. The patterns from voice agents shipped for plumbers, lawyers, and immigration consultants. The shortest path from “add voice to our product” on a Monday to a working agent answering calls by the weekend.

Touch the book to rotate it

Out now. $9.99 USD on Kindle, paperback available.

Author

Mahimai Raja J

Mahimai builds and ships production voice agents on LiveKit. He has shipped voice agents for plumbers, lawyers, and immigration consultants, and learned every failure mode in this book the hard way. The handbook is the field guide he wishes existed when he started.

In the community

About the book

The brief “add voice to our product” lands on a backend developer’s desk every week now. What follows is usually three weekends of trial and error: which STT, which LLM, which TTS, how to put it on a phone line, why the agent confirms data the caller never confirmed, why the call drops after thirty seconds, why the build cost six times what the demo predicted. The handbook collapses those weekends into one read.

You start at the LiveKit quickstart and end with a voice agent answering calls for your business: a phone number, a web button, a mobile feature, and the deployment shape that does not crash under load. Nine chapters. Three surfaces.

Every code block runs. Every claim has a number behind it.

Who this is for

You’re a backend developer. Someone on your team (maybe you) decided the product needs voice. You’re comfortable with async Python. You’ve shipped LLM-based features before. You’re trying to figure out what voice actually means in code, infrastructure, and provider choices.

If that’s you, this book skips your first three weekends.

If you’ve never opened a terminal, this isn’t the book for you, and I’ll tell you that up front.

What you’ll know how to do

What this book isn’t

This isn’t a guide to training your own STT or TTS models. It isn’t a deep dive on WebRTC internals. It isn’t about chat-style LLM products with optional voice. It isn’t a comparison of every framework: it picks LiveKit and goes deep.

If you want any of those, you want a different book.

What is inside

IntroductionFree sample

PART I: FOUNDATIONS

  • 01The Voice Agent Stack on LiveKitFree sample
  • 02Your First Agent

PART II: BUILDING REAL AGENTS

  • 03Prompts for Voice
  • 04Function Tools, State, and Handoffs
  • 05Latency, Interruptions, and Turn-Taking
  • 06Common Patterns
  • 07Surfaces

PART III: PRODUCTION

  • 08Testing with Synthetic Callers
  • 09Deployment and Observability

Conclusion

APPENDICES

  • AChoosing Your StackSTT, LLM, TTS, noise cancellation. May 2026 snapshot.
  • BThe Rest of LiveKitEgress, Ingress, avatars, telephony, the supervisor pattern.

The 800 ms budget

StageBudget
Network, user to server~50 ms
VAD and turn detection~100 ms
STT finalization~150 ms
LLM time to first token~300 ms
TTS time to first byte~150 ms
Network, server to user~50 ms
Total~800 ms

Each term is a real, measurable budget. A 600 ms end-to-end feels instant; 1500 ms feels broken. Every 100 ms you shave from the LLM line is directly audible.

A sample of the writing

The agent that mishears "drain cleaning" as "drainage exorcism" is funny. The agent that won't stop talking when a frustrated caller is trying to be heard is the one that costs you the customer.

From the Conclusion

Markdown formatting that helps in chat becomes literal "asterisk asterisk" noise in voice.

From the Introduction

Opus is the reviewer; Haiku is the receptionist.

Get your copy

Get the Book

Out now. $9.99 on Kindle.
Pick your local Amazon store.

Buy on Amazon USAAmazon Canada

Prefer print? Buy the paperback on Amazon USA →

Questions

Will the book be updated as the field changes?
Yes. Voice AI moves fast. Updates through 2026 are included in your Kindle purchase. KDP pushes new versions automatically.
Why LiveKit specifically, not Pipecat or building from scratch?
Three reasons: the SIP integration is the cleanest I've found for getting real phone numbers wired up, the client SDKs cover every surface I've needed (web, iOS, Android, Flutter, Unity), and the AgentSession abstraction matches how I actually think about voice agents. If you end up choosing Pipecat instead, you'll still ship a good agent.
Do I need a LiveKit Cloud account to follow along?
Yes. The free tier is enough for the whole book.
Is there a paperback?
Yes. Both the Kindle edition ($9.99) and the paperback ($24.99 CAD, local pricing varies by store) are available now on Amazon.
Will there be a video course?
Not yet. If enough readers ask for one, maybe.
What if it doesn't help me?
Email me. Refunds go through Amazon's standard 7-day policy on Kindle. If you bought the book and it didn't earn its $9.99, I want to know why.

Read the Sample Today

Free PDF in your inbox.