PitchHut logo
AI-powered finance categorization that keeps your data local and secure.
Pitch

NumbyAI is a local-first AI finance categorizer that automatically processes bank statement CSVs, ensuring all data remains on your machine. By leveraging a powerful local language model (Ollama), it categorizes transactions, identifies spending patterns, and provides insights without any cloud reliance or data sharing.

Description

NumbyAI: An AI-Powered Personal Finance Transaction Categorizer

NumbyAI offers a cutting-edge solution for personal finance management by providing a local-first transaction categorizer that ensures your sensitive bank data remains private on your machine. This fully-featured application utilizes FastAPI and React combined with the power of Ollama's local language model to efficiently categorize transactions from CSV bank statements without any need for cloud services or subscriptions.

Key Features

  • Automatic CSV Detection:
    Upload any bank statement in CSV format, and NumbyAI will automatically detect the layout and categorize each transaction using sophisticated algorithms. The system is designed to handle the messy realities of real-world bank exports, ensuring minimal configuration is required.
  • Comprehensive Dashboard:
    Track your spending across different categories, monitor your budgets, and analyze cash flow trends from a single, easy-to-use interface.
  • Rule Advisor:
    The built-in Rule Advisor analyzes past transaction patterns and suggests reusable categorization rules to facilitate future uploads.
  • Multi-Currency and Bank Support:
    NumbyAI can handle a variety of bank formats and currencies, accommodating various international standards for date and number formats.

How It Works

  1. Upload CSV: Drop your bank statement where NumbyAI quickly auto-detects metadata rows and column layouts.
  2. Initial Categorization: A rule engine runs first, using saved patterns for instant categorization of known transactions.
  3. Leveraging Local LLM: For any ambiguous transactions, the system sends batches to the local LLM (Ollama) for categorization. Confident results are automatically committed while uncertain ones enter a review queue for your evaluation.
  4. Review and Suggestions: Easily review flagged transactions, create new categorization rules, and analyze historical data for inconsistencies and improvements.

Supported Formats

NumbyAI supports various bank formats across the globe, including:

  • Chase (US): 7-column format with post date and category information.
  • Barclays (UK): Inflow/outflow split columns for comprehensive financial tracking.
  • Revolut (Multi-Currency): Handles various export formats from one of the leading international banking applications.

Privacy-First Approach

NumbyAI ensures that privacy is prioritized with zero telemetry, no external API calls, and complete local processing. Your financial information is never shared or transmitted online.

Visualization and Analysis

Integration of visualization tools enables users to monitor spending across different categories and track monthly budgets against actual expenditure through detailed charts and reports.

Simple Setup

Deploying NumbyAI is straightforward, with support for Windows, macOS, and Linux. Users can utilize Docker for quick deployment or set it up manually via Python and Node.js.

For further insights, refer to the NumbyAI Demo on YouTube and explore how NumbyAI can transform personal finance management while keeping data secure.

0 comments

No comments yet.

Sign in to be the first to comment.