Attention to Detail · Being Human · Shipping, Not Talking · Attention to Detail · Being Human · Shipping, Not Talking · Attention to Detail · Being Human · Shipping, Not Talking · Attention to Detail · Being Human · Shipping, Not Talking ·
Less Manual Work · No Lost Enquiries · Data That Updates Itself · Before and After · Clean Handover · One Person in Charge · Less Manual Work · No Lost Enquiries · Data That Updates Itself · Before and After · Clean Handover · One Person in Charge ·
CLICK ANYWHERE TO SKIP
RYBKIN ALEKSEI
I automate whatyou do by hand
Websites, bots and CRM integrations. They pay for themselves in your team’s hours.
To get straight to the point, tell me in your first message:
what needs building, or what is still done by hand;
the first result you need, and who will use it;
what it has to connect to: a site, CRM, 1C, spreadsheets;
timeline and budget, if you already know them.
Case studies
Client project
Roton
medical supplies store
+12%
search traffic three months after the move
20 h
a month freed for a manager by the 1C exchange, by the client’s estimate
Problem
The site had been hacked, a page took 4 seconds to load, and a manager updated the products on it by hand.
Solution
Moved the shop to a clean build, set up the 1C exchange and ordering through a chat.
Result
A page loads in 1.1 s instead of 4.1 — from “poor” on Google’s scale to “good”. The 1,100 products update from 1C on their own, orders come in through the chat, and search positions were back within three weeks.
to analyse a typical contract; the protocol takes seconds
₽10–20
per 20-page contract with Claude Sonnet, October 2026 prices
Problem
Buyers send the supplier contracts drawn up in their own favour: uncapped penalties, deductions from payment, the right to walk away without paying for goods already made. Each one has to be read and answered with a protocol of disagreements.
Solution
Built a web service: a manager uploads the contract, and AI finds the risks for the supplier, explains them in plain words and proposes new wording for the clauses.
Result
A contract is analysed in minutes, and the protocol of disagreements comes out in Word and PDF, with the signature and stamp, at the press of a button. From the brief to a working system on the server took about a week.
spent on the AI model’s API: Claude Code scores the vacancies and writes the letters on the user’s subscription
5
sources in one panel: hh.ru, Telegram channels, company sites, tenders and studios
Problem
Job hunting on hh.ru, Russia’s largest job board, takes hours: a few searches return hundreds of vacancies, every letter is written from scratch, and applications and employers’ replies have to be tracked by hand.
Solution
Built a panel: it finds vacancies through the official hh.ru API and filters them by rules, while Claude Code scores each one and writes a letter from the facts in the CV.
Result
Vacancies arrive with a score and a letter, every application is on one board, and invitations and rejections come to Telegram. A letter that lacks facts won’t go out until they are filled in.
a year saved on payroll: one person does the content work of three, estimated from salaries
Problem
A hundred-plus posts a month need three people — a content manager, a designer and a marketer — and one post took 4 hours.
Solution
Built a pipeline on the Claude API: it writes the copy from the brand’s templates and orders the images itself, and a human picks and polishes.
Result
A post takes 20 minutes instead of 4 hours, and 56% of drafts ship unedited: 500+ posts in five months from one person.
EMCD Tech
IT and fintech
14 h
a week went on writing call minutes by hand; now it is 5 minutes of proofreading after a call
Problem
Call minutes were written by hand, agreements got lost between calls, and reports were pieced together by copying between spreadsheets — with errors.
Solution
Brought in call transcription with an automatic summary and a list of decisions, and set up spreadsheet syncing and auto-filled reports.
Result
Minutes are ready after 5 minutes of proofreading, and finding a past discussion takes seconds. Twelve manual steps and the copying errors left the weekly cycle.
Madnomad Design
design and marketing agency
4
ad layouts a day instead of one
Problem
A designer put the property developer’s banners together by hand — one layout a day, and every adaptation to a new placement was another round.
Solution
Moved the graphics production to generative models in the brand’s style, and requirements sign-off to checklists and briefs.
Result
A layout takes a quarter of the time, revisions and new formats need no round with a designer, and post-delivery rework is down 60%.
I do on my own what usually takes a team. AI speeds up the work, while the architecture and the checks stay with me: a result is accepted only after automated tests and review. One person answers for everything from the task to the launch, not a chain ofcontractors.
