AI avatars for kiosks and conversation training

AvatarTalk AI avatar, click to chat
Talk to me!

Deploy conversational AI avatars in public venues like airports, museums and entertainment spaces. Train your teams with realistic video practice before they face real customers, guests, patients or employees.

  • 🖥️ Full offline appliance option
  • 🔒 Private by design
  • 🎭 Realistic avatar practice
  • 📊 Scoring and coaching notes

Two ways to use AvatarTalk

A focused platform for AI avatars in public venues and measurable conversation practice for customer-facing teams. Available in the cloud, on-premise or embedded on-device. Use hosted online generation and our API, or run everything locally as an offline appliance.

Offline AI Kiosks

A complete AI avatar appliance for kiosks, machines and public venues that runs locally without relying on cloud inference.

  • • On-device ASR, VAD, LLM, TTS and avatar runtime
  • • Low-latency conversations without internet dependency
  • • Private by design, with conversations kept on the appliance
  • • Built for airports, entertainment venues, museums and public venue operators
Explore AI Kiosks

AI Training Simulations

Video-first AI avatar practice that helps teams prepare for difficult conversations with customers, guests, patients and employees.

  • • Hospitality, retail, support, sales, leadership and healthcare scenarios
  • • Realistic pushback, interruptions, mood shifts and escalation
  • • Manager assignments, scoring and coaching notes
  • • Built for roles where presence, tone and de-escalation matter

Built for real-world deployment

Offline runtime

Run conversations locally on dedicated hardware.

Privacy posture

Avoid sending sensitive conversations to cloud inference.

Manager visibility

Track practice, scores, coaching notes and repeat attempts.

Deployment help

Support from pilot to production rollout.

Full offline deployment

AvatarTalk can run the full conversation loop locally for kiosks in airports, museums, entertainment venues and other public locations.

  • On-device runtime: ASR, VAD, LLM, TTS and avatar rendering on dedicated hardware.
  • Private by design: sensitive conversations can stay inside the appliance.
  • Hardware options: runs directly on DGX Spark, with lower-cost edge GPU deployments such as NVIDIA L4.

Online API and demos

For teams that prefer hosted workflows, AvatarTalk also supports online avatar generation, streaming examples and API integrations.

Discuss your AvatarTalk deployment

Tell us if you are exploring offline kiosks, training simulations or a hosted deployment.