Locx - Intelligent Automation
OCR and automated data extraction from 13,000 rental contracts with Generative AI.
Duração: 6 weeksEquipe: 2 developers

Principais Features
- OCR
- Generative AI
- LLMs
- Prompt Engineering
O Desafio
Processing and extracting structured data from over 13,000 rental contracts in various formats (scanned PDFs). The manual process would take months and be subject to human errors. A solution was needed that could handle different layouts and document qualities.
A Solução
We implemented an automated processing pipeline using advanced OCR combined with Large Language Models (LLMs) for intelligent data extraction. The solution uses specialized prompt engineering to ensure extraction accuracy, automatic data validation, and structured storage in the database. All processing occurs on the backend without the need for a graphical interface.
Resultados
- 13,000 contracts processed automatically
- 97% accuracy in data extraction after optimization
- 90% reduction in processing time vs. manual method
- Estimated savings of thousands of hours of manual work
Tecnologias
- Python
- OpenAI GPT-4
- OCR
- FastAPI
- PostgreSQL
- Docker