Concept for an AI video for Yandex Station Midi

Yandex Station Midi is a compact smart speaker that sets the mood. Music, a voice assistant, and smart home scenarios — all packed into a single minimalist cube. It’s not just a device. It’s the hub for controlling the vibe in your home.

At the agency, we created a full-length commercial entirely using AI — without a single day of filming. No cameras, no set, no production team in the traditional sense. Instead, we used a controlled generation system where every element of the scene is defined and controlled: the character, their appearance and movements, the location, lighting, composition and camera movement. We don’t «generate images» — we assemble the scene like a construction set and direct it within the AI. Everything can be changed: from the position of a hand to the lighting mood — without reshoots or losses.

This provides speed, flexibility and control that simply do not exist in traditional production. Essentially, it is a new production logic, where an idea immediately becomes a shot.

Instead of casting, we brought in our own creator. We digitised the appearance: we compiled a set of head angles to capture the face and avoid ‘floating’ in the generation.

We put together the outfit separately: clothes and accessories — all as a modular system. This gave us a consistent character who could be placed in any scene without losing their identity.

We didn’t have to look for a flat either — we put it together ourselves. A modern Scandinavian interior: light, air, minimalism, subtle accents. Every item is a deliberate choice: the sofa, the table, the plants, the lighting. We set the scene so that the speaker was the focal point of the composition and the action.

Find the five cats
 ↓

The fifth cat is you

The script is as simple as possible, and therefore effective. One continuous shot with no cuts. The character walks into the flat, flops down on the sofa, reaches for the speaker and presses the button. At that moment, the space transforms — from a quiet flat into a home club. The lighting changes, movement and energy appear; the sound literally «switches on» a new reality.

The final highlight is the product. The speaker makes a clean 180-degree turn. And this is crucial: it’s not a 3D render. It’s the same AI-generated imagery, but with control over form, lighting and materials, just like in a product shoot.

To create a vertical version of the video, we didn’t simply crop the footage to 9:16. We expanded it. Using AI, we filled in the space beyond the original frame and recomposed the shot. The result is a fully-fledged vertical version, not a cropped adaptation.

No animals were harmed in the making of this video.