Hishob (हिशोब): A Private Money Dashboard With a Local AI That Finds Where It Goes
Описание и применение
Hishob — локальный дашборд для домашнего бюджета, использующий LLM Qwen3.5‑122B‑A10B и Neo4j для объединения банковских транзакций, счетов и чеков в единую графовую модель, позволяющую точно отслеживать расходы и экономию.
Задача
Household finances are fragmented across multiple accounts and documents, making it hard to see where money actually goes and to identify unnecessary spending.
Результат
The system now provides a single, accurate view of household finances, automatically detecting over‑payments, idle cash, and mortgage interest effects, while keeping all data local and private.
Рабочий процесс
- 1. Pull bank transactions once a week by hand via the CDR feed.
- 2. Forward bills and statements to a dedicated Gmail mailbox; filters label them by type.
- 3. Nightly job reads labelled emails and Nextcloud uploads, parses PDFs, and files them.
- 4. All data is stored in Neo4j; relationships link accounts, transactions, bills, and payments.
- 5. Overnight AI analyst runs rule‑based detectors and generates insights.
- 6. Dashboard displays aggregated spending, interest, and savings across all accounts.
Доказательства
The first series ran everything on a 6 GB mini PC with a 3B model, and part 2 ended with "a better LLM requires better hardware".
Only two things reach outside the house, and both only read: the bank feed (through an accredited CDR intermediary) and Gmail over IMAP.
Everything lands in a Neo4j knowledge graph, so a debit, the bill it paid and the line on that bill are connected, not just sitting in separate tables.
- Дата записи: 2026-10-18T07:41:09.106Z