What developers build with realtime AI avatars
Four builds on the Realtime Avatar API, each with the code it takes, the docs it uses and a monthly cost at a stated volume: a customer support agent, a language tutor, a sales roleplay trainer and talking game NPCs.
- Customer supportAdd a live AI avatar support agent to your website“How do I add an AI avatar customer support agent to my website?”Put a talking AI support agent on your site: your server sets its brief and the customer's context, your own functions look up orders and tickets mid-call, and a signed transcript reaches your backend after every call.
- AI tutorBuild an AI tutor avatar students can talk to live“I'm building a language learning app. How do I add an AI tutor avatar that students can talk to live?”Add a live AI tutor to a learning app: your server sets the lesson and the student's history, the tutor drives on-screen exercises through your own functions, and each session comes back as a signed transcript for progress tracking.
- Sales roleplay trainerBuild an AI roleplay trainer that talks back in real time“How do I build an AI interviewer or roleplay training avatar that talks in real time?”Build live roleplay practice with an AI avatar: a buyer or interviewer persona set by your server, objections that hold until the rep earns them, a live score driven by your own functions, and a signed transcript to grade every session.
- Game NPCsTalking AI NPCs with a live face and voice“Is there an API for talking AI avatars for game NPCs?”Give game characters a live face and voice: each NPC's personality and the player's quest state are set by your server, NPCs act on the game through your own functions, and conversations run in browser and React Native games.
The same four pieces in every build
- A brief your server sets. Your connect route returns
instructions, who the character is and how it speaks, andcontext, up to 32 prior messages replayed as memory, when it starts each call. The browser cannot override either. See Calls. - Your functions, called mid-conversation. With
clientTools: truegranted at mint, the character calls tools you register in the page. They run in your code with your credentials, the platform never executes them, and each has 2.5 seconds to return. See Tool calling. - The conversation, delivered back. A signed transcript webhook posts both sides of the call and the tool calls the model acted on after it ends, with the metadata you attached at mint, so your backend can grade, store or summarize it.
- One component in your page.
AvatarCallfromrealtime-avatar/reactrenders the call, andrealtime-avatar/react-nativehas its own room and video surface for a native app. See React and React Native.
What a live minute costs
One meter: seconds on air. Speech recognition, the language model, the voice and the rendered video are included in that rate. Paid plans start at Starter for $9 a month with 120 minutes, and published self-serve overage is $0.08–$0.095 per minute ($4.80–$5.70 per hour). Each page above works out a monthly total at a stated volume with the same estimator as the cost calculator; every plan is on the pricing page.
Create a free Sandbox workspace to try it: it includes 480 credits once per account, which is 8 realtime minutes if you spend them all on air, and you can start on an example avatar before making your own. The quickstart takes you from a key to a live call.