Cases in production

Trucks arrived at the plant with no prior order: queues at the weighbridge, crossed phone calls and an operation that depended on the memory of whoever was at reception that day.

A Telegram agent now coordinates the unloading: drivers announce their arrival from the cab, the agent assigns them a slot based on material and available dock, and the plant sees the queue in real time.

The result: less waiting at the weighbridge, zero coordination calls and a complete record of every unloading without anyone writing anything down.

The salon's phone rang while clients were being attended: half-written appointments, unrecorded changes and lost slots.

A WhatsApp agent manages the calendar with real availability: it books, changes and cancels appointments in natural conversation. Including the hard part — time zones, daylight-saving changes and weekend edge cases that used to break appointments.

The salon stopped losing bookings outside opening hours and the calendar stays consistent without manual work.

Every morning someone transcribed the orders arriving by email by hand: SKUs, quantities, variants. Slow, and with errors that ended up in the delivery round.

A pipeline reads incoming email, an LLM parses each order and validates it against the catalogue, and the order is created in Odoo via JSON-RPC. Doubtful SKUs are flagged for review instead of guessed.

From inbox to registered order, hands-off, with exceptions flagged so a person only decides where it adds value.

The pickup operation was coordinated over phone calls and WhatsApp groups: nobody had the full picture and proof-of-service photos arrived late or not at all.

A single flow puts it in order: a Kanban board for the office, a PWA for drivers with their route of the day, a photo inbox over Telegram from the truck and automatic WhatsApp alerts to the client.

Every pickup is documented on the spot, and the office sees the state of the whole fleet at a glance.