Invoice traking & Accounting automation

Built a multi-workflow automation system for an accounting/tax advisory client to eliminate manual invoice and bank statement processing.

Invoice Pipeline: Captures invoices automatically via Gmail and Google Drive triggers, validates file type, and uses an AI agent to classify whether each document is a true invoice. Valid invoices are parsed by an LLM-based extractor pulling structured data (vendor, amount, date, line items), then logged into a Google Sheets database. A second AI agent cross-references extracted invoices against existing records to detect matches — automatically updating matched rows, renaming and routing matched files to a tax-advisor handoff folder, or flagging unmatched entries for review.

Bank Statement (CAMT) Pipeline: Ingests CAMT bank statement files via upload or download trigger, extracts and splits individual transactions using AI parsing, formats the data, and appends it to a dedicated transactions database in Google Sheets.

Result: A fully automated intake-to-database pipeline spanning invoices and bank transactions, reducing manual bookkeeping work and creating a clean, structured handoff point for tax advisors — built end-to-end in n8n with OpenAI-powered extraction and classification at every decision point.