Case study · Content & SEO
An automated blog for a car workshop: articles from social posts and fresh search data
A car workshop films its own work. The system turns every such post into an SEO article on the website, and every day finds topics for new articles in fresh Google and Yandex search data. All the owner has to do is click “Approve”.
12 days
from kickoff to a working system
32 articles
published on the client's site in the first 10 days
9 of 32
on topics from fresh search demand
$0
in model costs: a private AI server writes the articles
The challenge
The client is a car workshop with a Bitrix website and active Telegram and Instagram accounts: real job videos go up there every week. But the blog — the part of the site that actually brings in search traffic — was rarely updated: there was no one with the time to write it.
- Content exists, but it lives only on social media and brings in no search traffic
- A copywriter doesn't know the technical details, and the mechanics don't have time to write
- No visibility into what people are actually searching for right now
- Box-checking articles hurt SEO more than they help
How the automation works
- 1
New social post
Telegram and Instagram: video, photos and caption pulled in full
- 2
AI watches the video
the car, the service, before-and-after measurements, the best frames
- 3
An AI writer drafts the article
target query, structure, FAQ, links to services, frames from the video
- 4
Checked: SEO and facts
an SEO checklist and an AI critic — invented details, errors and repetition get sent back
- 5
Owner approves
approve, edit, leave a comment, or swap the image
- 6
Published on the site
scheduled in Bitrix, old articles cross-linked, sitemap updated
Topics from the freshest search engine data
A second line runs in parallel. Every day the system pulls fresh statistics from Google Search Console and Yandex Webmaster: which queries the site already appears for, how many impressions and clicks it gets, and at what position. Yandex Wordstat adds search volume, and Google and Yandex search suggestions show what else people look for around those queries. New article topics come out of this data.
- Finds queries where the site is already close to page one but isn't capturing traffic yet — an article on these pays off fastest
- Ranks topics by real numbers — impressions, position and search volume — with those figures shown next to every suggested topic
- Once a topic is approved, AI researches it and grounds it in the company's own work — the article draws only on matching completed jobs
- The numbers behind every suggestion come only from the search tools — the model can't make them up
- Google Search Console
- Yandex Webmaster
- Yandex Wordstat
- Google search suggestions
- Yandex search suggestions
What the system can do
Articles built on real jobs
Car, problem, what was done and how it turned out — with the numbers from the video, not generic filler.
Images pulled from your videos
The system picks the best frames and a cover image itself, and adds photos from the post. Any image can be regenerated from a description.
An independent AI critic
Checks facts against the source and catches invented details, technical errors and repetition. The owner's comments are binding: the critic checks that every one was addressed.
Publishing without spikes
A schedule, a weekly cap and a buffer of finished articles: hundreds of old posts get worked through gradually, without the bursts search engines penalize.
Any model for any role
Claude, OpenAI, Gemini or your own model on a dedicated server — roles are switched in settings, with no rework.
Built to be reliable
Access keys are stored encrypted, and a backup is created on a separate server every night.
Integrations
- Telegram
- Bitrix
- Google Search Console
- Yandex Webmaster
- Yandex Wordstat
- Google and Yandex search suggestions
- Anthropic Claude
- OpenAI
- Google Gemini
- Your own model on a dedicated server
- Backups to a separate server
Results
- The first article went live on day three
- 32 articles in 10 days, 9 of them on topics from fresh search demand
- The blog updates regularly with no copywriter: the owner spends a few minutes a day on it
- 800–2,000-word articles with an FAQ and links to services — built on real jobs and real numbers
- Model cost is $0 per article: a private AI server writes and checks them
Who it's for
- Car workshops and detailing studios
- Repair and renovation
- Clinics and dental practices
- Construction companies
- Manufacturers
- Any business that can put its work on video
Questions and answers
Won't the articles sound like they were written by a robot?
No. Every article is built around a specific job: the vehicle, the problem, what was done and what came of it. The critic strips out generic filler and invented details.
Where do topics for new articles come from?
From fresh data in Google Search Console, Yandex Webmaster and Wordstat. The system finds queries where the site is already close to page one and suggests topics with the numbers behind them. You decide what actually gets written.
Can articles be edited before they're published?
Yes: the text is edited right in the panel, and you can leave a comment for the writer or swap an image with one click. Nothing publishes without your approval — unless you switch on fully automatic mode yourself.
Does this work with platforms other than Bitrix?
It's ready to go for Bitrix. Other CMS platforms are connected on request.
What do you need from us?
Read access to your social accounts, website access to install the publishing module, half an hour to profile your company — who you are, who you write for, what you can't promise — and a few minutes a day to review articles.
Want the same for your business?
Send a request — in 30 minutes we'll walk through your process and show you what AI can take off your plate.
Free consultation