All cases
01 Case · Content automation

The math of $60k saved a year

You can’t trim a content team a little at a time — a copywriter, designer, video editor and researcher cost like a small apartment every month. I replaced them with one Telegram bot that ships content around the clock for a dollar a day.

n8nDockerFastAPITelegram APIFFmpegLLM orchestration
$5,000/mo
the team used to cost
$1/day
the bot costs now
$60k+
saved per year
24/7
shipping, no pauses
n8n automation graph
Under the hood — n8n orchestration: dozens of nodes from generation to publishing.

The stakes

Content production is bottlenecked by people: every post passes through a chain of hands, and any of them can get sick, take a vacation, or stretch a review into three days. A team like that runs ~$5,000/mo and still doesn’t scale.

The brief was simple: close the whole loop — from idea to a finished file — and make it run with no people and almost no money.

The move

One Telegram bot as the only interface. Under the hood — n8n orchestration: request routing, LLM calls, visual generation, FFmpeg video rendering and publishing. All packed into Docker on a dirt-cheap VPS.

From the chat you pick a format — content plan, carousel, script, edit, long-read or an events calendar — and the bot drives it to a result, asking for confirmation only at the key steps.

What it does

Long-reads: writes articles with real links to research — it parses and cites sources, returns ready HTML.

Carousels: generates a slide-by-slide script and assembles the visuals — “one project, different packaging”.

Video: analyses the source, transcribes speech, cuts subtitles and edits a vertical Reel via FFmpeg.

Events: parses award and festival deadlines and sends urgent reminders with submission links.

The result

A content pipeline that doesn’t get sick, take vacations, or wait on reviews. Cost — about $1 a day vs. $5,000/mo for a team: over $60,000 saved a year, with faster output.

Gallery

The bot’s main menu
A single Telegram interface — pick a format, the bot delivers a file.
Carousel script generation
A slide-by-slide carousel script — with text approval.
A generated long-read with sources
A long-read with real links to research — parsed and cited.
Reel editing via FFmpeg
A vertical Reel edit: speech, subtitles, stitching via FFmpeg.
Award deadline parsing
Event parsing: urgent award deadlines with submission links.

Want the same — or stronger?

I’ll build a system like this for your task.