Case study
AI-supported contract analysis for construction projects
Development of a locally executable AI system for analysing extensive construction contract packages. The goal was to make commercially and technically critical information across 200 to 300 pages of contractual material easier to locate and review in a structured manner.
Initial challenge
In practice, contract packages often consist of the main agreement, specifications, annexes, meeting minutes and additional contractual documents. For estimating, risk analysis and project execution, however, only selected topic areas are truly critical. Manual review is time-consuming, repetitive and prone to oversight.
Solution approach
The solution combines document processing for digital and scanned PDFs, OCR for image-based documents, a hybrid retrieval architecture covering both lexical and semantic search, and topic-specific extraction logic with plausibility and confidence mechanisms. Outputs are transformed into a review-friendly format including context, source locations and an Excel-based review layer.
Outcome
Even in an early project phase, the system was able to identify key clauses on contract value, retention, daywork rates, guarantees, parent company guarantees, insurance and payment terms with strong quality. Scanned documents also produced usable results despite typical OCR-related degradation.
Business impact
The solution materially reduces manual screening effort, improves systematic coverage of critical clauses and creates a traceable basis for commercial and technical review decisions. Its modular structure also provides a scalable foundation for additional document types and review topics.