The whole business lived in three messengers and the owner's head. Now it is one system.

What it was like

• Client requests arrived as calls and voice messages, worker contacts were scattered across three messengers and several accounts.
• The database was a set of spreadsheets with no phone numbers and no documents.
• An urgent request at 8pm meant calling every available worker one by one: an hour, sometimes two.
• Workers are paid daily, clients pay weekly or monthly — and all of it was held in the owner's head.

Where it got difficult

• This was not "install a CRM": requests come in free form — "three people for unloading at the arena tomorrow at nine" — and have to become structured data automatically.
• The workers are not people who enjoy complex interfaces; signing up had to be as simple as possible.
• The database holds identity documents for site passes, which in Russia means consent, local storage and encryption by law.
• The budget was not corporate: the project needed stages where the next level is built on top of the previous one.

What was built

• One bot working in two messengers at once, Max and Telegram: everyone uses what they already have.
• Requests by voice or text: the system transcribes, turns it into a structured request and shows it for confirmation.
• Four ways to distribute a shift: post to the channel, target selected workers, priority workers first with automatic fallback, direct assignment.
• Workers register themselves: the document is just a photo, the data is recognised automatically.
• The bot keeps availability current, sends shift reminders, collects confirmations and starts looking for a replacement when someone drops out.
• All the money in one place: payouts, incoming payments, debts, taxes, profit.
• A web admin panel as the single control centre instead of spreadsheets and three messengers.
• Personal data handled by the book: consent in the bot, data and models on Russian servers, encrypted documents, access log.

Result

An urgent evening request no longer means hours on the phone: the bot sends it to available workers and collects responses. The architecture is laid out for the next stage — a marketplace — to be built on top.