The copy machine that follows the SEO markup
Every text needs a brief: phrases, headings, length. Usually an in-house writer turns it into copy. Instead: a machine that keeps every text lined up with the SEO markup itself.
Outcome
- Similar texts no longer wait in the queue for a free copywriter
- New products reach the site sooner: a text is ready in minutes, not days
- Fewer rewrites: key phrases and headings from the brief are already in place
- Nothing new for the team to learn: it runs through Telegram and ordinary Word files
The problem
In shops with a large catalogue, search specialists already write briefs for texts on a regular basis: key phrases, heading structure, length. What is missing is someone who sticks to that markup every time, so the shop looks for an in-house copywriter.
The work is routine but not mechanical: a person gets tired, shifts a heading or drops a key phrase. A machine does not get tired: it writes the hundredth text to the same brief as the first.
What I built
A text in minutes, not days. The specialist sends the brief in Telegram as the same Word file they already prepare for a copywriter, and gets a finished .docx back within minutes.
Instructions get carried out, not copied. The file is parsed down to its formatting: the code sees the highlight colour of every fragment and hands the model the text and the instructions separately. A highlighted «make this an H2», «alt for the photo» or «turn this into a table» gets done and never lands in the text for the site.
Nothing gets lost. Links from the brief stay links, Meta Title and Meta Description come first, and both language versions come back in the same order.
When it fits
- Search specialists write the brief, but nobody writes the final text. Every brief waits in the queue for a free copywriter.
- The same format repeats hundreds of times. A person gets tired and loses a phrase or a heading somewhere.
- The brief is written in Word with colour highlights. They have to be carried out rather than copied into the text.
The last word stays with a person
The machine returns a file, it does not publish. The specialist reviews the finished .docx and moves it to the site, so the machine has no access to the shop at all: not the catalogue, not orders, not customer data.
Common questions
Who publishes the final text? A person. The machine sends the file back in Telegram, and the specialist moves it to the site after reviewing it.
What happens to instructions for the writer inside the brief? The machine carries them out: sets heading levels, builds tables, collects alt texts into a separate block. The instructions themselves never reach the finished text.
Stack
- Python
- LLM
- aiogram 3
- python-docx
- SEO
Got a similar task?
Describe it in two sentences. I'll tell you if it can be automated, and we'll set up a short call to go through the details.