Palpfinis
A personal finance autopilot for Brazil.
Working MVP · independent project · since June 2026
Most personal finance apps still ask people to record, fix and categorize their own spending. Palpfinis starts from the opposite idea: bring together the sources that already exist, organize what is unambiguous, and ask for attention only when there is a genuine exception.
What already works
Palpfinis's first product is ContasMensais, a private web application that started with my own monthly expenses and already works across mobile and desktop.
- tracks monthly expenses, payments and notes;
- keeps recurring expenses without duplicating or erasing historical occurrences;
- reads Ceará NFC-e receipts from the camera, an image, a QR Code URL or an access key;
- uses the fiscal purchase date to file the expense under the correct day and month;
- preserves merchant, subtotal, discount, total and every receipt line item;
- categorizes products, accepts corrections and learns rules per merchant and product;
- prevents the same NFC-e from being imported twice;
- persists data in Cloudflare D1 and requires a preview before any fiscal data is written.
Today, fiscal classification is local and deterministic: purchase data is not sent to an AI model. Dates, amounts, recurrence and deduplication remain the application's responsibility.
The next step
The plan is to turn the MVP into a consent-based financial view that brings together accounts and cards from multiple institutions through an authorized Brazilian Open Finance integration partner.
With transactions and receipts in one place, Palpfinis can match a card purchase to its NFC-e, explain what made up the amount, and prevent a receipt and its corresponding transaction from becoming two separate expenses.
Where Claude fits
Claude will handle the ambiguity that fixed rules cannot resolve well:
- interpreting abbreviated and noisy Portuguese merchant descriptions;
- normalizing products without changing the original fiscal description;
- suggesting matches across receipts, banks and cards with confidence levels;
- detecting unusual or potentially duplicated charges;
- explaining cash-flow changes in direct language;
- producing proactive alerts without making users open a dashboard every day.
AI suggests and explains. The backend calculates, validates and writes. Sensitive actions still require explicit user confirmation.
Goal
I want Palpfinis to be the financial app that asks for the least maintenance: instead of entering every expense, users review a short exception queue and get on with their lives.
Technology and stage
React · TypeScript · Cloudflare Pages Functions · Cloudflare D1 · Ceará NFC-e integration
The MVP is currently used privately by the founder. Open Finance integration and the Claude-powered layer belong to the next product phase and are not presented here as already launched features.