CoolFace
Apppublic

fullstack-overlord/Kognit_AI

sourceHugging Facemitupdated 6mo agoView on Hugging Face
1likes
App README

Sabi-Spend: Your Smart AI-Powered Accounting Assistant

Sabi-Spend is a high-performance, AI-powered accounting assistant designed to help Nigerian small business owners track sales and purchases, calculate profit and loss, give profitability advice and draft statement of account all in their own language—including English, Pidgin, Yoruba, Hausa, and Igbo.

Built by Ridwan Oyeniyi [[fullstack_overlord](https://github.com/fullstack-overlord)]

🚀 Live Demo

[Get started with Sabi-Spend]

✨ Key Features

  • —Multilingual Voice Support: Talk to Sabi-Spend in English or your local dialect (Yoruba, Hausa, Igbo and Pidgin).
  • —Automated Bookkeeping: Converts natural speech or text into structured accounting data instantl, removing the need for manual data entry.
  • —Official Statements & Insights: Generates insightful financial health reports and professional statements of account tailored to the business's specific performance.
  • —Empathic AI Accountant (High Reasoning): Powered by Gemini 3.1 Flash-Lite's "Thinking Mode." Sabi-Spend doesn't just calculate; it understands. It provides comfort during business setbacks and offers strategic, actionable advice for financial recovery.
  • —Interactive Financial Consulting: Speak to Sabi-Spend as you would to a human accountant. Users can ask complex questions about their business health, profit trends, Business wisdom/advice, financial health analysis or expense management and receive deep, nuanced analysis.
  • —Cloud Persistence: Data is synced to Hugging Face Datasets for 100% security and recovery, ensuring your records are never lost.
  • —Scalable Multi-Tenant Architecture: Built with a "Multi-Tenant" architecture; the app dynamically creates and manages separate, private ledger files for every unique Business Name, allowing the system to serve thousands of SMEs simultaneously.

🚧 Other Features (Coming Soon)

  • —I-O-U Tracker (Debt Management): Tracks who owes the business money and automated payment reminders. Allow users to record debts using an updated CSV structure with debtor_name and status (paid/debt).
  • —Bank-Ready PDF Statements: Converts messy CSV data into professional, "Bank-Ready" PDF documents using FPDF2. Perfect for business owners applying for loans or grants.
  • —AI Anomaly Detection: An "AI Safety Net" that flags potential entry mistakes (e.g., recording a ₦5,000 sale as ₦50,000) by comparing inputs against previously recorded business prices.
  • —Paper-to-Digital (OCR): Leveraging Gemini’s vision capabilities to turn a photo of a handwritten receipt or printed invoice into an automatic digital entry.
  • —Daily/Monthly PDF Reports: Users can prompt Sabi-Spend to generate the sales/expenses recorded on a specific day or for a specified month. Use Pandas to filter the CSV by date.

🛠️ Tech Stack

  • —Language: Python
  • —Brain: Google Gemini 3.1 Flash-Lite (via Google GenAI API)
  • —Interface: Gradio (optimized for mobile)
  • —Storage: Hugging Face Datasets (CSV/TXT)
  • —Deployment: Hugging Face Spaces

⚙️ Setup Instructions

  1. 1.Clone the repo: git clone https://github.com/fullstack-overlord/Sabi-Spend.git
  2. 2.Install dependencies: pip install -r requirements.txt
  3. 3.Setup environment variables in a .env file:
  4. 4.GEMINI_API_KEY
  5. 5.HF_TOKEN
  6. 6.Run the app: python app.py

🛡️ License

Distributed under the MIT License. See LICENSE for more information.

Built with grit during a 7-days (12-hours/day) development sprint for The 3MTT Knowledge Showcase.