Use cases

AI agent for accounts receivable control

The agent collects debt from 1C and CRM, buckets overdue invoices by age, warns about the cash gap and drafts a reminder for every late payer.

Receivables slip out of control not because customers refuse to pay but because the debt is noticed late. The shipment is posted in 1C, payment is due in forty-five days under the contract, the account manager remembers their own deal in Bitrix24, accounting sees the balance per counterparty — and nobody sees the whole picture: how much is outstanding, whose overdue has passed a month, what that does to next week's payment calendar. The question surfaces on the day the money is needed, and the work starts from zero.

The usual routine is a monthly export of the customer settlements ledger, overdue rows coloured by hand in Excel, and a mail-out to the sales team. By the time it lands half the data is stale: some invoices have been paid, some customers have gone deeply overdue. Reminders are written one by one in free form — ten debtors is half a day of work, so they get written for some customers and not for others, and rarely at the right moment. The opposite question, what you owe suppliers this week, lives in another file and belongs to another person.

The agent keeps that picture live. It takes settlements and payment terms from 1C, deals and owners from Bitrix24 and amoCRM, incoming money from the Tochka statement and documents from Diadoc, then splits the debt by counterparty and ageing bucket: under 30 days, 30 to 60, over 60. Alongside it computes the opposite flow — what you owe and when — and shows which day of the week the cash gap appears. For each debtor it drafts a personal reminder carrying the amount, the document numbers and the history of the conversation. Sending, and every legal step beyond it, stays a human decision: the agent does not write to your customer in your name without your go-ahead.

Common questions

How does the agent know a payment is overdue?

From the contract terms and the documents: shipment date, payment delay and incoming money per counterparty come out of 1C, and the fact of payment out of the bank statement. Debt counts as overdue once the due date has passed with no closing payment behind it.

Will it message the debtor itself?

It prepares the draft — with the amount, the delivery note numbers and a tone that fits how late the payment is. Sending needs your confirmation: a first gentle reminder can be automated, a conversation about a formal claim cannot.

Does it forecast the cash gap too?

Yes. The agent combines account balances, expected receipts by due date and the week's committed payments — taxes, payroll, suppliers — and warns you which day the money runs short, before it becomes a fact you learn at the bank.

Does debt data come from 1C or from the CRM?

From both, when both are connected. Amounts and terms are more reliable in 1C; the responsible manager and the conversation history live in Bitrix24 or amoCRM. The agent matches the counterparty across the two and shows the debt next to whoever owns it.