Real-time talking avatars for product tours.

One API to put a face on your product: it listens, answers from your documents, and acts through your tools.

ENfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to SophiaSophiaENyour guide to AvatarGo
ENfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to LucasLucasENsales and demos
ENfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to MayaMayaENcustomer support
ENfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to JamesJamesENinterview practice
中文first frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to MeiMei中文客服 · customer support
ENfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to OliverOliverENEnglish practice
ENfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to AmaraAmaraENclinic front desk
ESfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to DiegoDiegoESbookings and reservations
ENfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to RuthRuthENrole-play for training
日本語first frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to KenjiKenji日本語hotel concierge
FRfirst frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to CamilleCamilleFRvisite guidée · museum guide
한국어first frame 1.1 s · interrupt 21 msfirst frame 1.1 s
interrupt 21 ms
Talk to Min-junMin-jun한국어영어 선생님 · English for kids

Each one is a character: a face, a voice, what it knows and what it can do. Point at one, then talk to it.

I answer in about a second.

Hand me your documents. I'll learn them and find your users' answers.

One photo made me. One API puts me on your site.

Customer supportSalesLanguage tutoringTraining

Put a face on the conversations that matter.Scroll, or pick one. Each plays a short scripted call.

listening
Mayacustomer support · live
youHi, where's my order? It's 48213.
toollookup_order · #48213 ✓
MayaIt left the warehouse yesterday and should arrive Thursday. Want the tracking link?
youAnd can I return it if it doesn't fit?
toolknowledge · returns-policy.pdf ✓
MayaYes, within thirty days, free of charge, in the original packaging. I can send you the —
youSorry — can I talk to a person?
MayaOur team is in from nine to six. I've noted your order and the question, so they can pick it up first thing.

How it works.From a photo to a character on your site. No code until the last step.

  1. Step 1

    Add a face

    Upload a photo you have the rights to, or pick one of ours. Then choose how it moves: a Live avatar follows the voice with its head, eyes and mouth; a Photo avatar keeps the picture still and costs the least.

  2. Step 2

    Make it a character

    Give it a voice and a language, your documents to answer from, your tools to act with, and a greeting. Then talk to it right in the console.

  3. Step 3

    Put it to work

    Put it on your website or in your app.

    Read the quick start →

What you getEvery tile shows the thing itself. All of it from one API and one SDK.

One photo, one faceUpload a photo; it is checked, framed and ready to call.
Knowledge from your documentsPDFs, Word files, notes: it searches them mid-call and answers from what it finds.
Tools that call your APIsLook up an order, book a slot, open a page in your app.
Natural interruptionsCut in mid-sentence; it stops and listens, like a person.
Live captionsBoth sides, word by word, in the language spoken.
Type or talkUsers can type instead of speaking; the avatar still answers aloud.
Voices in many languagesPick a voice; it speaks the language your users do.
Pre-rendered videosA picture and script in, an MP4 with subtitles out. Talking head or half-body.AI video →
0:10
portrait
Your own audioAlready have a voice pipeline? Send the audio; we render the face and lip-sync to it.
sto answer, typically
msto stop when you interrupt
fpslive video, frame by frame
phototo start: upload it, pick a voice

Two calls to a live avatar.Your server trades an API key for a session token; the page joins with it. The key never reaches the browser.

Read the docs
Your server
// a session token for this user — your API key stays here
const res = await fetch("https://api.avatargo.ai/v1/auth/session-token", {
  method: "POST",
  headers: { authorization: `Bearer ${process.env.AVATARGO_API_KEY}`,
             "content-type": "application/json" },
  body: JSON.stringify({ characterId: "ch_…" }),
});
const { sessionToken } = await res.json();
Your pageJavaScript
// the avatar in <video id="avatar">, with the token from your server
import { createClient } from "@avatargo/sdk";
const client = createClient(sessionToken);
await client.streamToVideoElement("avatar");

Questions.

What do I need to start?

A photo, or one of our faces. Sign up free: your account starts with credit to try it, and you build and test your first character in the console without writing code.

Can I use my own face, or someone else’s?

Yes, if you have the rights to the picture. For someone else’s face you need their consent; you confirm it when you add the picture.

Which languages can it speak?

The languages of the voice you pick. It also hears which language a user speaks, so people can talk to it in their own.

Can it use my documents and my systems?

Yes. Give it your documents and it answers from them during the call. Give it tools and it can look things up or act through your APIs, such as checking an order or booking a meeting.

Can I bring my own voice or AI?

Yes. Use just the avatar: your server sends the speech audio, and the face says it, lip-synced.

How is it priced?

From one balance, billed by the second of each call. See pricing →

Build your first character.

A face, a voice, what it knows and what it can do. Then talk to it right in the console.

© 2026 AvatarGo