UA Message the founder
Case studies

Before and after in nine businesses.

What we built for clients, how it works and how fast it pays back. “Per client data”: the client gave us the number. “Estimate”: we calculated it, the assumptions are on the case page.

9 systems in productionpayback from 2 monthsevery number is labeled with its source: per client dataestimateCalculate the payback on your numbers (in Ukrainian)

We describe clients by type and publish names only with their consent.

  • ProcurementConnected systemsin production

    Forecast-based purchasing for restaurants and catering

    Client
    4 restaurants and catering; before the project, one kilogram of food in five went in the bin, ≈€170k a year
    What we built
    Takes sales from the tills of every location, forecasts demand for each product and drafts tomorrow's order itself; the manager approves it with one tap in Telegram, and at 07:45 the suppliers have their orders
    Status
    In production

    7.5%per client dataof purchases goes in the bin, was 21%: 2.8 times less

  • Customer chatsConnected systemsin production

    An AI salesperson on Instagram, from a comment to a Nova Poshta shipment

    Client
    Clothing store on Instagram, 17 products in ≈350 variants
    What we built
    Replies in messages and comments within a minute day and night, finds the product from a photo, checks stock in the CRM, creates the order and the shipping label, upsells; the owner confirms payment
    Status
    In production

    +15%per client datarevenue in the first 2 months with the agent

  • Customer chatsCallsin production

    AI front desk for a dental clinic chain

    Client
    Dental clinic chain: 3 clinics, 11 chairs; before the project, one patient in six missed the visit
    What we built
    24/7 booking in chats and by phone, reminders and reconfirmation, a waitlist for every cancellation; the administrator confirms complex bookings
    Status
    In production

    6%per client dataof patients miss their visit, was 17%: 2.8 times fewer

  • Customer chatsCallsin production

    Four AI agents for a car dealer hub

    Client
    Car hub importing from US and Korean auctions; before the project, 3% of requests ended in a sale and the first reply took 1–1.5 days
    What we built
    The agents message and call every request right away, add the client to the CRM without duplicates, find cars, guide the client until pickup and buy a “Buy Now” lot when the manager taps the button
    Status
    In production

    up to 3 minper client datafrom the request to the first reply, was 1–1.5 days

  • Finding clientsin production

    New business clients, and no letter goes out without a human

    Client
    Maker of corporate merch, sells to companies. Before the project the team worked through 2,000 companies in 5 months by hand
    What we built
    A profile of each company and a letter to the person who decides, sent after the manager's “ok”, conversations on Instagram and Telegram, prices, invoice and payment, reminders and an AI assistant for the whole business
    Status
    In production

    20–25%per client dataof people we wrote to for the first time reply, on average across all channels

  • Custom softwareCustomer chatsConnected systemsin production

    Custom CRM and an AI salesperson for an electronics store chain

    Client
    Chain of used electronics stores: sells to people and companies
    What we built
    A custom CRM with every channel and call in one place, each request routed to the right manager; the AI salesperson filters out random questions, closes simple sales and learns from the best managers
    Status
    In production

    +5–8%per client datamore sales from the same number of inquiries in the first month

  • Procurementin production

    Demand forecast and a ready order to manufacturers

    Client
    Veterinary and pharmacy products brand: up to 200 items, 2 contract manufacturers, 10 years of sales history
    What we built
    A forecast for each item and a ready order to the manufacturer within the stock budget, with an explanation and a shortage warning; the buyer confirms
    Status
    In production

    −22%per client dataless money sits in stock in the warehouse

  • Custom softwareConnected systemsin production

    Loan decisions in 40 seconds, with personal terms for everyone

    Client
    Non-bank lender: ≈40,000 applications a year; before the project, a specialist spent 1–2 days checking each one
    What we built
    Assessment on the lender's own loan history, document checks, a personal rate, amount and schedule; a specialist decides doubtful cases with a ready analysis
    Status
    In production

    40 sper client datafrom application to ready terms, was 1–2 days

  • Custom softwareConnected systemsin production

    A therapy booking platform with a catalog, packages, payment and calendar

    Client
    An online platform for psychological support: people choose a psychologist themselves and pay for sessions
    What we built
    A platform built from scratch: a quiz that matches a psychologist, booking of free time slots, card payment, session packages, accounts for clients, psychologists and admins, emails and analytics from click to payment
    Status
    In production

    2 minper client datafrom the button on the website to paying for a session

Next step

Pavlo, the founder, replies. The first call is free.

Message Pavlo on TelegramAll contacts
Page text for AI assistants (markdown)Open index.md
Loading markdown…
UA
Message Pavlo, the founder