Speaks your visitor's language.
Janice explains the Holobox in Dutch. She answers in English just as well.
IRIS AVA
Life size. Real.
Yours.
Interactive avatars of real people, built from 4K footage for Holoboxes, Holofans and portrait displays. In our cloud or on your own hardware, with your own language model.
Janice is listening
A Holobox is a holographic display. You see a person or a product as if they were really standing in front of you. There are four sizes. The 22 and The 43 are compact and fit on a counter. The 75 and The 86 are large, and in The 86 a person stands at life size. Only need one for a few days, for a trade fair or an event? Then you can rent a Holobox too. We deliver it and collect it again afterwards.
Een Holobox is een holografisch display. Je ziet iemand of een product levensecht, alsof het echt voor je staat. Er zijn vier formaten. The 22 en The 43 zijn compact en passen op een balie. The 75 en The 86 zijn groot, en in The 86 staat een persoon op ware grootte. Heb je hem maar een paar dagen nodig, voor een beurs of een evenement? Dan kun je een Holobox ook huren. Wij leveren hem en halen hem daarna weer op.
In the wild
Not a window.
A presence in the room.
Other avatars live in a browser tab. Ours stand in shops, lobbies and trade fair halls, where people walk up to them. These are real Holobox installations by Holoconnects, filmed on location.
Real installations with our clients' own content. IRIS AVA brings a conversational avatar to the same screens.
Wherever people ask
One avatar.
Every front desk.
Illustrations: Janice's real avatar placed in generated rooms.
4K, life size
Shot in 4K.
Down to every curl.
A person at full height, a metre from your visitor. So we start from a 4K portrait recording and keep it. Only the face is generated. Hair, hands, clothes and posture stay exactly as they were filmed.
- 2160 × 3840Native 4K portrait source, body and background
- 25 fpsFace animated live while she speaks
- 1 GPUAbout 32 GB runs the renderer, in our cloud or yours
Live conversations stream at 1080p today. The 4K live route, where the screen plays the 4K body and only the face is streamed, is in testing. We will show it to you on a screen before we promise it.
Every moment
Present. Even between
the answers.
The whole face speaks.
Eyes, lips and teeth move with every word.
Listening.
Breathing, waiting, looking at you.
No cut.
From listening to talking, seamlessly.
1
Only one thing is AI.
The face. Everything else is the real person, in 4K.
Portrait first
Made for the screens
people walk up to.
Most avatar platforms live in a browser window. IRIS AVA is made for standing, portrait screens in the real world, and works on the web as well.
Holobox
Life size in The 86. On a counter in The 22 and The 43. Buy or rent.
Holofans
A figure floating in mid air that stops people in their tracks.
Portrait displays
Kiosks, reception screens and showroom totems.
Your website
The same avatar, for visitors who start online.
IRIS OS
One OS.
Every screen.
Every Holobox, Holofan and display connects through IRIS OS. Switch a screen live between your content, a real colleague and your avatar, and manage every site from one place. The IRIS appliance runs it next to any screen and drives up to four 4K displays.
SPRING COLLECTIONYour campaign, on every screen.
IRIS AVA
Your infrastructure
Your building.
Your model.
A hotel, a bank and a ministry have different rules. So your avatar does not have to live in someone else's cloud.
01
Our cloud
We run the renderer and the conversation for you. The fastest way to start.
Pilot02
A dedicated GPU
Your own reserved graphics card, in the region you choose. Nothing shared.
Pilot03
On premises
The renderer on one NVIDIA GPU of about 32 GB in your server room, with IRIS at the screen. Nothing leaves the building.
In testingYour language model
Connect a local LLM.
Point the avatar's brain at a model you host yourself, through any OpenAI compatible endpoint such as Ollama or vLLM. Speech recognition already runs on our own GPU. Or use ours, answering only from the knowledge you approve.
- Speech recognitionWhisper, self hosted
- Language modelYour local LLM, or ours
- KnowledgeOnly what you approve
- Face and bodyYour 4K avatar
How it works
One recording.
Then it's yours.
-
Step 1
Record
A few calm minutes in the browser studio, or a 4K session with us. Listen, speak, listen.
Try the studio now -
Step 2
We build it
Our team prepares face, body and idle motion. You see the result and approve it before anything goes live.
Pilot -
Step 3
It answers
Connect your knowledge and your screens through IRIS OS. Your avatar goes to work.
Pilot
Use cases
Where a face
makes the difference.
Reception and wayfinding
Greets every visitor and explains where to go.
Showrooms and retail
Explains products in depth while staff focus on buyers.
Trade fairs and events
Draws people to your stand and answers all day.
Hotels and hospitality
A concierge in the lobby, day and night.
Banks and government
On premises, with a local model, for the strictest rules.
Museums and training
A guide or trainer who never tires of questions.
Trust
A person's face
deserves care.
Consent first
We only build an avatar of someone who gave explicit, written permission.
Your knowledge only
Answers come from content you approve. Anything else is declined.
People approve
Nothing goes live until you have seen it and said yes.
Yours to withdraw
Ask us to remove an avatar and we delete it, with its recording.
Is Janice a real person?
Yes. Janice recorded herself once, in 4K, and gave her permission to be used as an avatar. In the answer on this page her face is generated by our renderer and her body and hair are the original footage. The voice is a licensed synthetic voice; a customer can choose their own.
Is it really 4K?
The body and background are native 4K footage. The face is generated and composited in. Live conversations run at 1080p today; the 4K live route is in testing, and we will show it to you before we promise it.
Can it run on our own servers?
That is what we are building for customers who need it: the renderer on one NVIDIA GPU of about 32 GB in your server room, with IRIS at the screen. We are testing it now. In the pilot you can start in our cloud or on a dedicated GPU.
Can we use our own language model?
Yes. The avatar can think with a model you host yourself, through an OpenAI compatible endpoint such as Ollama or vLLM, so questions and answers stay inside your network.
How does it work with IRIS OS?
IRIS OS runs every Holobox, Holofan and display. Your avatar becomes one of the modes on a screen, next to your content and a live person, managed with the rest of your fleet.
Can visitors hold a live conversation?
Live conversations are what we are building and testing with our pilot customers. The answer on this page is a recording, so you can judge the image for yourself.
What does it cost?
Pilot terms are agreed per project, based on your screens, where it runs and how many conversations you expect.
Pilot programme
Be one of
the first faces.
We are onboarding a small group of pilot customers. Start with a recording today, or tell us where your avatar should live.