Malinski.AIClaimed by Official Representatives

Challenge: Supplier invoices arrived in a shared mailbox in every possible form - clean PDFs, printer scans and phone photos. For each one, someone had to open the attachment, identify the document type, retype the invoice number, tax IDs, dates and net, VAT and gross amounts into the ERP, and then check for typos. At 3-5 minutes per document and 80-120 invoices a month, this took 6-10 hours of manual work every month, and a single mistake in a tax ID or VAT amount could mean a correction later.

Solution: We built an AI tool that reads invoices from e-mail attachments, extracts the data and creates the document in the ERP without retyping. Documents that raise doubts go to a short review queue, where a person approves them.

Result: The accountant no longer retypes invoices and spends the time on checking and cost analysis instead.

Challenge: Weekly and monthly sales reports were built by hand: exporting tables from the ERP, copying them into Excel, building formulas, pivot tables and charts. It took hours, the data was partly out of date by the time management saw it, and copy errors distorted results. Only one person knew the exports, the steps and the formulas, so when that person was on leave, the report was late or not produced at all.

Solution: We automated the report: data is pulled from the ERP, processed and delivered as a ready report with a short AI commentary, without anyone assembling it by hand.

Result: The report no longer depends on one person, arrives on schedule and the team spends its time analysing the numbers instead of compiling them.

Challenge: An online retailer settled with a payment operator, two courier companies and a marketplace. Each sent reports in a different format, some only by e-mail. At month-end the accounting team spent hours matching payouts to invoices in the ERP, accounting for operator commissions, courier surcharges and corrections, and cash-on-delivery payments that arrive days after shipping - and still had unexplained differences.

Solution: We built a tool that automatically reconciles what each operator and courier settled with what the ERP shows, and lists only the differences.

Result: Month-end reconciliation runs on its own; the team explains only the flagged differences and can show which report and document each figure comes from.

Challenge: Company knowledge was spread across a 100+ page PDF FAQ, knowledge base articles and an internal wiki, each with a different structure and no common search. A new customer-service employee needed several minutes to answer a simple question; HR questions such as leave entitlement or medical-care rules meant searching long regulations. Some answers were known only to one experienced person.

Solution: We built an AI agent that answers employees' questions from the company's own documents.

Result: Answers are available immediately, also when the person who used to know them is away, and new employees get up to speed without searching through documents.