Private project · Money & administration
Finance & Home Operations
I use an assistant alongside a structured financial toolkit to organize the administration of a real home project. The work connects scattered documents and messages with a ledger that distinguishes an invoice, a payment and the budget category it belongs to.
Python · SQLite · PDF extraction · Gmail/IMAP · Google Drive · WhatsApp exports · CSV · Excel
Preserve the document and its context
Import tools collect files from Drive, invoice emails and WhatsApp chat exports. The WhatsApp importer connects each PDF with the surrounding conversation, including when it arrived and the adjacent explanation. It extracts candidate invoice fields into a reviewable CSV; scans without a usable text layer are flagged for manual description.
That separation matters: extracting a number from a document is a proposal, while adding it to the ledger is an explicit operation. The original document and its context remain available for checking.
A deterministic financial core
SQLite stores categories, budgets, invoices and payments. Views calculate gross amounts, balances due, payment status and budget utilization. The reporting layer distinguishes issued invoices from actual payments and exposes cost by category.
Excel exports retain formulas so the result remains useful for manual review and further editing. I chose explicit records and deterministic calculations for the financial core, with assistant-driven work around those tools.
Extending the workflow
I am also developing a separate Gmail-to-expense-sheet integration. Its next step is defining the target schema and review process before implementing the extraction flow. Across both projects, the aim is to turn repetitive administration into a sequence of inspectable operations while keeping the financial record understandable.