S-08 / Content & growth automationServices

Content & growth automation

The moment the person who posts gets busy, the channel goes quiet for two weeks. Now briefs turn into scheduled posts every day, in your voice and in three languages.

Book an audit Message on Telegram ↗

S-08.1Problem

What it looks like today

We have a Telegram channel and an Instagram page, and both look abandoned. Our SMM person left in spring; the manager posts when she remembers. One post takes an hour: find a photo, ask someone for a Russian caption, translate it into Uzbek ourselves, forget English. The Uzum listings are up to date, the channel is not.

Agencies quote us a junior salary for twelve posts a month and a report, and the Uzbek captions still read like Google Translate. Then someone told us clients now ask ChatGPT and Yandex for the best dental clinic near them — and we are not in the answer. We do not want a bigger agency. We want a pipeline that runs every day, in our voice, and a page that search engines and AI can find.

S-08.2Outcomes

What changes in your business

A content pipeline living in your Telegram: one-line brief → generated image or short clip + captions in uz/ru/en → two-button approval → scheduled post to Telegram and Instagram
Target: 14 posts a week across your channels at under $1 API cost per asset and under 10 minutes from brief to queued post — numbers we measure on our own channels first
Nothing publishes without a person pressing approve; every asset is logged with prompt, model and cost, so a post has a price, not a guess
Uzbek captions in correct Latin script with ʻ and ʼ, Russian without Uzbek calques, English for the export audience — one brand-voice document behind all three
Product posts straight from your data: a 1C price list or a Uzum catalogue export becomes a week of posts in one evening
Trilingual SEO/GEO for your site: per-language pages, hreflang, JSON-LD, llms.txt and a monthly report on what Google, Yandex and AI answers show for your 20 target queries

S-08.3How it works

Three steps, in this order.

01Step 1
Channel audit
45 minutes on a call and read-only access to your Telegram channel, Instagram and site analytics. In 5 working days we come back with: what you posted in the last 90 days and what it cost, the 20 queries you should be found for, a draft of the brand-voice document and the pilot scope.
02Step 2
Pilot in 30 days
Week 1: the bot is connected to your channels and the brand voice is signed off. Weeks 2–4: daily publishing with approval by your person; the first trilingual landing page or article goes live with hreflang and JSON-LD. Day 30: the digest with posts, cost per asset, reactions and the first search positions.
03Step 3
Retainer
At an agreed monthly rate the pipeline keeps running: briefs, drafts, scheduling, the weekly digest and two new SEO pages a month; the extended package adds short video, three languages on every channel and AI-answer (GEO) work. Cancel monthly; the bot, prompts and content stay yours.

S-08.4Deliverables

What you sign off on

Content pipeline bot (grammY + Claude API + Supabase) connected to your Telegram channels and Instagram, running in your own accounts
Brand-voice document (uz/ru/en, 2–3 pages): tone, vocabulary, forbidden calques, formats per channel
Editorial calendar for 30 days with formats, languages and posting times; refilled every week
Approval flow in your ops chat with a daily spend cap you set
Weekly digest: posts published, cost per asset, reactions, what to repeat
SEO/GEO set for the site: 20 target queries per language, hreflang, JSON-LD, per-locale OG images, llms.txt; a monthly position report for Google and Yandex plus a check of what AI answers say about you
Runbook (3–5 pages) and admin rights: the team can run, pause or hand the pipeline to someone else without us
Asset library: every generated image and clip with its prompt and licence note

S-08.5Price

What it costs.

By agreement

Exact quote after the free 30-minute audit.

S-08.6Proof

We run what we build.

Fact 01
The pipeline we sell is the one that runs our own channels: the agency, AcademyOS, the football side of the work. 14 posts a week, under $1 per asset and under 10 minutes brief-to-post are targets until 30 days of publishing are logged — then they become measured, right here.
Fact 02
This site is the SEO/GEO reference: three languages, hreflang and canonical on every page, JSON-LD, per-locale OG images and llms.txt — the same set we ship for your site.
Fact 03
57.9% of surveyed Uzbek businesses say they have integrated AI; the talent shortage is one of the four barriers named (UNDP, Feb 2026). A pipeline replaces the designer, copywriter and translator you cannot hire — not the person who approves.

S-08.7Questions

Questions we get on the first call.

Is this only for shops, or does it fit a manufacturer too?

Both, but for different reasons. For a shop, content sells directly; for a textile, food or construction-materials manufacturer it works on trust and on distributors — technical pages, certificates, production news, the material the sales team forwards to a client. The pipeline is the same; the topics and the metrics are not.

Where do our data and content live, and who owns the channels?

Channels and accounts stay yours: you keep admin rights, the bot is added as a poster, tokens live in your own Vercel and Supabase projects, which we hand over. Briefs and drafts go through business-tier AI accounts where prompts are not used for model training. Your clients' personal data never enters the pipeline — it only needs product data and your brand voice.

Can it write proper Uzbek, and who approves?

Captions are generated in Latin Uzbek with ʻ and ʼ, checked against your brand-voice document, then a person in your team presses approve — nothing publishes otherwise. In the first two weeks we proofread every Uzbek and Russian draft ourselves; corrections go back into the prompts, so week four needs fewer edits than week one. Target: under 10% of drafts sent back by day 30.

How fast do we see results — and can it use our product data?

Posting starts in week 2 of the pilot; the day-30 digest shows posts, cost and reactions. Search is slower: Google and Yandex usually need 8–12 weeks to reflect new pages, and AI answers follow the pages they can read — so SEO/GEO is a 90-day line, not a 30-day one. Product data: yes — a 1C price-list export, a Google Sheet or a Uzum catalogue export becomes product posts; when a price changes at the source, the next post uses the new one.

What happens after the pilot, and can we take the pipeline in-house?

Three options: a monthly retainer at an agreed rate (we run it), a hand-over (you run it — the runbook, the bot, the prompts and the Supabase project are yours, plus 2 hours of training for your person), or a pause. There is no lock-in: the code written for you is yours under the contract; only our internal skill packs stay ours, and you can buy those separately.

S-08.8Next step

Start with the free 30-minute audit.

Five questions, one call. You leave with the workflow to automate first and a cost estimate.