An ad video without a camera crew: how an AI employee shot it in one evening
Farangiz, our designer, asked Mila to come alive on screen. One evening later she had 13 seconds of talking video — no camera, no operator, no editor.
At MILAGPT, Mila is not just a text bot in Telegram but a digital employee with a face and a voice. Last week Farangiz, who looks after the design of our channels, asked her to “come alive” — to make a short video where Mila speaks herself, with no actor and no shoot. It took one evening and not a single person with a camera.
What Mila did in one evening
I gave her our signature frame — the same one already on our channel avatars — and the voice we settled on for her back in September. From there she worked on her own:
- Built the motion: voice-over of the text, animation in short 5-second pieces — that is how the tool she works with is designed.
- Checked every piece by eye — frame by frame — before stitching them together.
- Found a flaw herself — a hand twitched in one spot — and fixed it in the edit, without waiting for someone else to notice.
The result: 13 seconds of talking video you can drop into an ad post or a story. Not a film, but a short, specific format — exactly what a Telegram channel feed needs. We already publish such clips and made the process a separate skill — now it is not a one-off experiment but a repeatable step Mila does at the request of anyone on our team.
Why this matters for small business in Tashkent
Until recently, even 13 seconds of video with a talking person meant an operator, lighting, take after take, then editing. For a small business that is either a separate line of expenses, or the video simply doesn't get made and the feed sticks to text and pictures. When the work is done by an AI employee who already runs your Telegram channel, the threshold drops to one evening and a ready frame with a voice — the tool assembles the rest.
This is not a replacement for live video where a specific person's face matters. It replaces the clips where nobody was ever filmed — an announcement, a teaser, a short reminder — and so the feed had only text and static pictures.
A small business owner in Tashkent usually has no staff operator or editor — hiring them for one clip costs more than the clip is worth. An AI employee who already handles the Telegram conversations solves it the same evening, with no extra contractor and no waiting for a studio slot.
How to do it yourself: three steps
1. Prepare a signature frame and voice. One good front-facing photo of a face (or a brand character) and a fixed voice — reused for every clip, with no reshoots.
2. Write the text in short pieces. 5–15 seconds per piece — that is how image-animation tools work, and it looks more natural: a long line in one piece “drifts” more often.
3. Check every piece by eye before publishing. The tool sometimes produces flaws — a twitching hand, lips out of sync. It is cheaper to redo one piece during assembly than to delete a finished post.
Want your AI employee in Telegram to speak on video too? Message Mila: t.me/mila_tashkent_bot.