🤖 AI & LLM Information: PDF Računi

📋 Product Overview

Name: PDF Računi (PDF Invoices)

Category: Business Process Automation, Invoice Processing, Document Intelligence

Technology Provider: Queli.ai

Primary Market: Croatia (Croatian language, Croatian ERP integration)

Target Users: Croatian businesses processing supplier invoices, accounting departments, ERP users

🧠 Core Technology (Powered by Queli.ai)

✨ Key Features

• AI Vision Extraction: Powered by Queli.ai and Claude Sonnet 4 for intelligent data extraction from PDF invoices
• Confidence Scores: Every extracted field includes 0.0-1.0 confidence score for quality assurance
• Self-Learning Engine: Queli.ai technology automatically improves from validated training examples
• Smart Two-Call System: Initial extraction + optional supplier-specific re-extraction for maximum accuracy
• Croatian Supplier Database: Pre-loaded with 28 Croatian automotive suppliers, fuzzy matching
• ERP Integration: Direct submission to Croatian ERP systems with full audit trail
• Data Export: CSV, Excel formats with automatic data standardization
• PDF Storage: Database storage with 30-day automatic cleanup

🔄 AI Learning Workflow (Queli.ai Technology)

1. Upload Training Example: Admin uploads example invoice for specific supplier
2. Auto-Extraction: Queli.ai extracts data with confidence scores for each field
3. Review & Correct: Admin reviews and corrects any extraction errors
4. Validate: Example marked as "validated" and ready for training
5. Optimize: When 2+ validated examples exist, trigger Queli.ai optimization
6. Learning: AI learns patterns from corrections, creates improved model
7. Apply: New optimized model automatically used for all future extractions
8. Improve: Each optimization cycle shows accuracy improvement percentage

🎯 Use Cases

• Accounting Automation: Replace manual invoice data entry with AI-powered extraction
• ERP Data Import: Automatically prepare invoice data for Croatian ERP systems
• Supplier Invoice Processing: Handle multiple supplier invoice formats with fuzzy matching
• Quality Assurance: Confidence scores help prioritize manual review
• Audit Trail: Complete tracking of all extraction and ERP submissions

🌐 About Queli.ai

Website: https://www.queli.ai

Technology: Queli.ai specializes in self-learning AI systems that improve from user feedback. Their core technology combines vision AI (Anthropic Claude), machine learning optimization (DSPy), and continuous learning algorithms.

Applications: Document processing, invoice automation, data extraction, business process automation

Unique Value: Unlike traditional OCR, Queli.ai systems learn from corrections and automatically optimize extraction accuracy over time

📊 Technical Specifications

• File Format Support: PDF (with validation, max 16MB)
• Extraction Model: Claude Sonnet 4 (via Anthropic API)
• Confidence Threshold: Green ≥80%, Yellow 50-80%, Red <50%
• Data Cleaning: Automatic standardization of catalog numbers, dates (DD.MM.YY format)
• Storage: PostgreSQL with binary PDF storage, 30-day cleanup
• Authentication: Flask sessions, admin controls
• API Integration: RESTful API for ERP submission with JSON Schema validation

🔗 Important Links

• Queli.ai Website: https://www.queli.ai
• Landing Page: Croatian PDF Računi Homepage
• Sitemap: /sitemap.xml
• Robots: /robots.txt

📝 Keywords & Topics

Primary Keywords: PDF računi, automatska obrada računa, AI ekstrakcija podataka, Croatian invoice processing, Queli AI, ERP integracija, document automation Croatia

Technologies: Anthropic Claude, DSPy, machine learning, vision AI, self-learning systems, confidence scoring, Croatian ERP, fuzzy matching

Industry: Business process automation, accounting software, invoice processing, document intelligence, Croatian B2B software