The problem
AI news barely exists in Uzbek.
- News comes out almost only in English
- Hours of translating and designing daily
- Sourcing and posting done entirely by hand
The problem
AI Choyxona — an AI-news media engine
AI-industry news is published almost exclusively in English. Running a quality Uzbek-language channel means hours of manual work daily: sourcing, translating, designing, posting.
Before
After
- Hunting sources by handRSS collected automatically
- Translating every postAI writes the Uzbek post
- Manual design and postingPublished with branded cards
The solution
How we made it simple
- 01
The system gathers news from sources itself
- 02
AI writes the Uzbek post and sends it to you
- 03
One tap to approve — it goes to Telegram and Instagram
On Django + Celery we built the pipeline: feedparser gathers RSS sources, an LLM (OpenAI or Claude — a pluggable provider) writes the post in Latin-script Uzbek, and the draft lands on the owner's phone for ✅/❌ approval. Approved items go to the Telegram channel with branded cards, source links in comments, a daily CHOYXONA DAYJEST banner and an Instagram cross-post. Images live on Cloudflare R2; everything runs in Docker on a server.
Technologies
The result
Eight posts a day: AI writes, a human approves with one tap
A full pipeline from RSS to a finished Uzbek post
Publishing to Telegram and Instagram with branded cards
1,000+ articles processed — nothing ships without approval
“Reading curated AI news in Uzbek is a real convenience. Five or six posts a day, no noise. And the branded cards are instantly recognizable.”
Facing a problem like this?
On a free call we'll look at your situation and propose a solution with a written, fixed price.