The Full Digital Human Has Left the Uncanny Valley
Ojin, Berlin's Real-Time GenAI company, launches Portrait*, Presence*, and Voice*: three real-time models for agents you forget you're talking to. Meet Lou today at talktolou.ojin.ai.
Quick facts
- Who
- Ojin, Berlin's Real-Time GenAI company
- What
- Launch of Portrait*, Presence*, and Voice*. Three real-time models for digital humans.
- When
- Monday, 4 May 2026
- Where
- ojin.ai · talktolou.ojin.ai
- Price
- Speech-to-video from $0.05/min · $10 free credits for new accounts
The news
Ojin launches Portrait*, Presence*, and Voice* today: three real-time models that take the digital human across the uncanny valley. Demo: talktolou.ojin.ai.
Press release
BERLIN, May 4, 2026. Digital humans have always had a tell. Today Ojin is launching three real-time models that finally drop it: Portrait*, Presence*, and Voice*. Together they let anyone build a human agent with a face, a body, and a voice that holds up in live conversation. To celebrate, Ojin built the most uncanny demo it could think of: meet Lou, a devil now in residence at talktolou.ojin.ai. Lou is Ojin's playable proof that the full digital human has left the uncanny valley.
["CEO quote on the breakthrough goes here. 1–2 sentences in a confident founder voice. Suggested angle: what it means that the full digital human has finally crossed the uncanny valley, why this is happening at Ojin, and what it unlocks for the people building on the platform."]
— [CEO Name], [CEO Title], Ojin
The three models
Each one handles a single layer of the digital human, and they are designed to compose.
Ojin Portrait*
The head-and-shoulder model, built for the profile-picture frame. From a single image and a voice track, it generates a face that talks, listens, and reacts in real time, with low enough latency to hold a live conversation.
Ojin Presence*
The full-body model. The same one-shot input produces a full-figure agent that sits, stands, gestures, and moves through a scene without breaking the illusion.
Ojin Voice*
The voice layer, with multilingual synthesis, voice cloning, and voice design built in. It pairs the face and body with a voice that actually sounds like the person they belong to, in whatever language the conversation needs.
Meet Lou
Lou is a loser middle-management demon. He lives in your browser, he talks back, and he has thoughts (mostly about meetings and his failure in hell). He's lousy at standup, weak on the agenda, fond of low-stakes office mischief, and the last to leave any Slack thread. He shares an office with humans, and yes, he's a handful.
Lou at a glance
- Species
- Devil (specifically: loser middle-management demon)
- Habitat
- Your browser
- Strengths
- Office mischief, Slack thread persistence, opinions
- Weaknesses
- Standup, the agenda, basic competence in hell
- Built on
- Ojin Portrait* + Ojin Voice*
- Talk to him
- talktolou.ojin.ai
Customers
Some of the world's largest enterprise brands have built on Ojin, including:
"An inspiring partner for us, and technologically one of the best."
— Stefan Ponikva, VP Communication and Experience, BMW
"It's a new way to present our brand, an adventure that's deeply experiential and aspirational."
— Emmanuel Rousson, VP of E-commerce, Clinique
"The best enabler and partner to think of. Everyone thinking of building anything with realtime generative AI, absolutely go for it."
— Oliver Lange, Head of H&M Beyond
Available today
The platform is live at ojin.ai. Speech-to-video starts at $0.05 per minute, up to twenty times more cost-efficient than the previous generation of comparable models, with $10 in free credits for new accounts. The API is framework-agnostic and works with Pipecat, LiveKit Agents, or any custom pipeline.
Compliance & hosting
| Hosting | EU-hosted |
| Data protection | GDPR-compliant |
| AI governance | EU AI Act-compliant |
| MENA | PDPL-compliant (Saudi Arabia) |
| Enterprise | SSO & SCIM |
About Ojin
Ojin is Berlin's Real-Time GenAI company. The Ojin platform powers human agents with a face, a body, and a voice that hold up in live conversation, EU-hosted and available today at ojin.ai.
Asset downloads
Visual assets are organised below. All Ojin logos are pre-cleared for editorial use.