Day 1 at @AgoraIO's #iRTE2026 ✅
We brought Spatius real-time avatars to the floor and let people just talk to them. No script, no pre-recorded answers.
If you’re here, come talk with our live avatar and the team!
And that’s a wrap on @NVIDIA Apsara Conference 2026!
From creating digital selves to exploring real-world applications of AI avatars, it was exciting to connect with builders, businesses, and innovators shaping the future of human-AI interaction.
A big thank you to everyone who stopped by, tried Spatius, and shared their ideas — and to NVDIA for bringing us together!
Try it yourself: https://t.co/wi5RGSC8xG. Building with avatars? Get in touch.
And that’s a wrap on @NVIDIA Apsara Conference 2026!
From creating digital selves to exploring real-world applications of AI avatars, it was exciting to connect with builders, businesses, and innovators shaping the future of human-AI interaction.
A big thank you to everyone who stopped by, tried Spatius, and shared their ideas — and to NVDIA for bringing us together!
Try it yourself: https://t.co/wi5RGSC8xG. Building with avatars? Get in touch.
I put my late grandpa inside a Harry Potter–style magic photo on my desktop 👴 Click the picture and he comes to life, chatting with me in Cantonese in real time. It almost feels like he never left.
🛠 Built with:
@spatius_avatar : real-time digital human, rendered on-device, lip-synced to speech
@livekit : real-time voice
@deepseek_ai : the brain (his personality and memories are just two text files)
@Azure : Cantonese voice
Electron: desktop widget
@claudeai : Opus 5.5
When my husband saw it, he's excited and and wants to make one of his late grandpa to keep his 80-year-old grandma company. He also wants to make one of himself, so he can "work with me" every day 😆
So I've open-sourced it. Make one for someone you love 💛 https://t.co/iEx4zMJG6a
I put my late grandpa inside a Harry Potter–style magic photo on my desktop 👴 Click the picture and he comes to life, chatting with me in Cantonese in real time. It almost feels like he never left.
🛠 Built with:
@spatius_avatar : real-time digital human, rendered on-device, lip-synced to speech
@livekit : real-time voice
@deepseek_ai : the brain (his personality and memories are just two text files)
@Azure : Cantonese voice
Electron: desktop widget
@claudeai : Opus 5.5
When my husband saw it, he's excited and and wants to make one of his late grandpa to keep his 80-year-old grandma company. He also wants to make one of himself, so he can "work with me" every day 😆
So I've open-sourced it. Make one for someone you love 💛 https://t.co/iEx4zMJG6a
We’re bringing Spatius AI to the @nvidia Inception Pavilion at Apsara Conference 2026 in Hangzhou! 🚀
Want your own AI avatar? Visit our booth, join our community, and get free avatar creation credits!
📍 Hall 4, Inception Pavilion
📅 Sep 22–24
Are you building an AI app that could use a human face? Come say hi!
Hello from Hangzhou 👋 We’re OFFICIALLY on-site at @NVIDIA Apsara Conference 2026 today!!!
What if you could walk into a booth, have a real conversation with an AI avatar — and then leave with a digital version of yourself?
That’s what we’re bringing to Apsara with Spatius.
Come experience our real-time AI avatars in action, explore how they can power interactive experiences across education, banking, livestreaming, and more — and claim complimentary credits to capture and create your own avatar on-site.
Because the fun part isn’t just watching an AI avatar talk. It’s seeing what happens when you become part of the experience.
📍 We'd love to meet you at Hall 4, NVIDIA Inception Pavilion
Putting an AI teacher into a learning device isn’t about adding a bigger GPU.
The harder problem is making a real-time avatar run clearly, responsively, and reliably on the devices students actually use.
That’s what Spatius is doing with TAL Education.
With an edge-cloud architecture, Spatius enables:
• 1080p real-time avatars on mainstream education hardware
• Lightweight motion-data transmission instead of full-video streaming
• Lower compute and bandwidth requirements
• Real-time AI interactions, not prerecorded video
The AI handles understanding, reasoning, and teaching.
Spatius turns those responses into a digital teacher that can appear, speak, and interact in real time.
The real benchmark isn’t whether an AI avatar looks impressive in a demo.
It’s whether it can run reliably, every day, at scale.
https://t.co/4uKFzpPvws
What does an AI avatar look like when it moves beyond a standalone demo?
We created this scenario-based demo to show how Spatius avatars can fit into different product experiences:
🎙️ A virtual host engaging with audiences in a live room
🏦 A banking assistant answering common customer questions
📚 An AI tutor delivering interactive lessons and real-time feedback
💬 An AI companion enabling more natural conversations
Each scenario brings together a digital avatar, real-time voice, conversational AI, and a purpose-built interface.
The goal is not simply to make an avatar talk. It is to explore how digital humans can become a meaningful part of real products and user journeys.
Four scenarios. One adaptable avatar platform.
🔗 Explore the scenario demo on GitHub: https://t.co/E6hHgb6msc
🌐 Learn more about Spatius: https://t.co/xBh0mNx2Up
Which scenario would you build first?
#AIAvatar #DigitalHuman #ConversationalAI #RealTimeAI #CustomerExperience #EdTech
Everyone can build an AI avatar demo.
The real challenge is shipping one to thousands or millions of users.
Cloud GPU rendering makes avatars expensive to scale — every conversation becomes another expensive video stream.
Spatius uses edge-cloud hybrid rendering to reduce bandwidth and compute costs while keeping real-time, photorealistic interaction.
$0.007/min. Built for production, not prototypes.
https://t.co/OYu5TlILNz