Services
Documents, tenders, reports, a department’s daily routine. I work out how things run today and build the whole tool, from prototype to production.
Main focus
Invoices, transfer documents, acts and PDF scans → checked data in SAP, 1C or a spreadsheet.
Crawling platforms and sources, a relevance filter, alerts.
Spreadsheets with automation, small web apps, ETL and dashboards.
Web, mobile or hardware, from an idea to a working demo.
The AS-IS and the root causes: where exactly hours, money and accuracy are lost.
A working tool on your real documents, not a slide deck.
We measure hours, errors and payback, and decide what goes into production.
Roles, logs, monitoring, documentation. It runs without me.
Document AI · in production
−60 h / month
Checking invoices, transfer documents (UPD) and acts before payment: recognition, LLM extraction and strict checks instead of manual work.
Agents · in production
10+ platforms → 1
An overnight crawl of tender platforms, an LLM relevance filter, hybrid search and matching against the client’s inventory register.
Document AI · published
1,000 CVs < 1 h
An LLM scores CVs against the job’s weighted criteria, removes duplicate candidates and returns a ranked Excel file with its reasoning.
Tell me what eats the hours: what comes in, how much of it, and where it should end up. I’ll answer with how I’d approach it.