Call-center talking head
Katie, live from the front desk
A talking-head agent for call centers, intake, and concierge — face, voice, emotion, and a route to a human. This is Katie Fisher, Virgent's public AI.
Katie Fisher
Embodied intake agent — listen, speak, express, hand off. Same pattern as a 24/7 front desk.
Live conversation
Same Katie as the corner chat. Click Talk to speak; she transcribes on a pause, then answers in cached voice chunks.
What this demonstrates: Katie Fisher on a live video-call surface: she listens on the microphone, speaks with a real voice, shows emotion, and can hand you to a human — the same pattern as a call-center or front-desk agent.
How it relates to your business: Put a face on the brand's AI. Healthcare intake, bank concierge, hospitality, and admissions convert better when the first contact has presence — not only a text box.
Pain it solves: Text chatbots feel transactional. High-empathy front doors need a face and a voice, and staffing humans 24/7 does not scale.
Who it is for: Customer experience, healthcare, financial services, hospitality, and education teams where the front door changes conversion or trust.
- Healthcare intake that greets patients, captures the reason for the visit, and routes to the right clinic.
- Bank concierge that answers routine questions and warm-hands a human advisor for the complex ones.
- University admissions assistant on voice, around the clock, with a live talking head.
Technical Architecture
Same Katie, embodied interface
Preset chips still use cached cards and cached voice. Spoken questions are transcribed, then typed and spoken questions go through the same site agent as the corner chat: RAG, multi-model routing, and AG-UI actions.
Embodied Avatar Agent — speak or type, then voice in chunks
┌──────────────┐ preset chips ┌─────────────────────┐
│ Visitor │ ───────────────► │ Cached response │
│ orb + voice │ │ cards + ElevenLabs │
└──────┬───────┘ │ + viseme timeline │
│ Talk (mic) └──────────┬──────────┘
▼ │
┌─────────────────────┐ │
│ /api/avatar-stt │ │
│ Scribe v2 / Whisper │ │
└──────────┬──────────┘ │
│ transcript / type │
▼ │
┌─────────────────────┐ │
│ /api/site-agent │ │
│ (same as corner │ │
│ Katie chat) │ │
└──────────┬──────────┘ │
│ text shown first │
▼ ▼
┌─────────────────────┐ ┌────────────────────┐
│ Answer in chat now │ │ Sentence TTS queue │
│ + action buttons │ │ play 1, prefetch 2+│
│ Pause / rewind / │◄───────────│ cache + visemes │
│ replay from cache │ └────────────────────┘
└──────────┬──────────┘
▼
┌─────────────────────┐ book / handoff
│ Viseme orb + PIP │ ──────────────────► /contact
└─────────────────────┘Want this on your front desk?
We build voice and video agents that greet, qualify, and escalate — with a human in the loop when it matters.
Ready for your own embodied agent?
Custom embodied agents with a voice, a face that actually articulates, and a route to a human.
