ElevenLabs embed
Basic coffee-barista embed with ElevenLabs Agents as speech-to-speech — pipe agent PCM into createUtterance.
View example → Live demoexamples.liforma.ai
Production-quality starter apps for adding intelligent animated avatars, speech, and real-time AI experiences to your product.
Using ElevenLabs / OpenAI Realtime / Gemini / Deepgram / LiveKit? Add a Liforma avatar →
Try a Live demo on *.examples.liforma.ai, or clone and run ./start for the gallery on localhost:4000 plus demos on 4001–4010. Deepgram / Gemini need a local WebSocket proxy
(hosted pages explain how).
Already on ElevenLabs, OpenAI Realtime, Gemini Live, Deepgram, or LiveKit? Pipe vendor audio into a Liforma avatar with these embeds.
Basic coffee-barista embed with ElevenLabs Agents as speech-to-speech — pipe agent PCM into createUtterance.
View example → Live demoBasic coffee-barista embed with OpenAI Realtime as speech-to-speech — pipe PCM + transcript into createUtterance.
View example → Live demoBasic coffee-barista embed with Deepgram Voice Agent as speech-to-speech — pipe PCM + transcript into createUtterance via a WS proxy.
View example → Hosted noticeBasic coffee-barista embed with LiveKit remote agent audio — mint a token and bridge tracks via createUtterance (+ transcript).
View example → Live demoBasic coffee-barista embed with Gemini Live as speech-to-speech — pipe PCM + transcript into createUtterance via a WS proxy.
View example → Hosted noticeHello-world experience embed: one experience id and one Experience component. No custom speak or listen UI.
View example → Live demoCorner launcher: collapsed thumb until click, then an in-page overlay conversation. Light until expand.
View example → Live demoLesson-based tutor app with an “I am learning” control that sets Experience learningLocale (Spanish by default; also French, German, Japanese).
View example → Live demoMarvely-style scripted turns: presenter mode, speak() for tutor lines, manual Start/Stop listening, and host-side feedback.
View example → Live demoType text and press Enter — the experience speaks it. Toggle enqueue vs interrupt on speak().
View example → Live demoSvelteKit, vanilla HTML, Next.js, and React (Vite) are available today. Vue, Nuxt, Astro, and Angular are planned — the web component also works in those stacks meanwhile.
`<Experience />` from `@liforma/client/svelte` — full lesson UI with TypeScript and Svelte 5.
View framework →AvailableCDN script + `<liforma-experience>` — copy-paste friendly for any site.
View framework →Available`@liforma/client/react` plus optional App Router session route helper.
View framework →AvailableClient-side React with `@liforma/client/react` — no Next.js server.
View framework →Phase 3Second major ecosystem with broad international adoption.
View framework →Phase 3Vue production framework for full-stack apps.
View framework →Phase 3Content sites, marketing pages, and docs where experiences embed well.
View framework →Phase 4Enterprise coverage for large structured codebases.
View framework →