Client
Ministry of Economical Affairs & KWINK Groep seeking to analyze complaint procedures across public tenders.
Industry
Government / Public Sector
Service
Process Automation, Document Analysis, Data Processing, LLM Integration
Tech
Python, LLM, Excel, API’s
Challenge
The client required a large-scale analysis of 150 public tenders comprising approximately 15,000 documents. The goal was to analyze how complaints were handled during tender execution. Due to the project’s scope and the detailed answers needed for each tender, manual execution was not feasible.
We took on the
following challenges:
- Extract all necessary documentation from the government’s API
- Develop an efficient method to analyze all documents
- Create an effective approach to obtain accurate answers to specific questions
- Present answers in a format suitable for further analysis and aggregation
Solution
Document Extraction
Developed automated systems to extract documents from government APIs, processing over 15,000 documents across 150 public tenders.
Smart Pre-selection
Implemented keyword-based filtering to identify relevant documents for detailed analysis.
LLM Integration
Utilized advanced language models for case-by-case analysis, ensuring accurate and consistent results.
Structured output
Created standardized output formats in Excel with clear yes/no and multiple-choice answers, backed by source document references.
Impact
Significant Reduction in Project Throughput Time
The automation reduced manual processing by 90%, shortening the project duration from months to weeks. This efficiency gain freed up resources for higher-value activities like in-depth analysis and decision-making.
Enhanced Consistency and Reliability
Automated detection and classification minimized human error. The consistent output format (yes/no and multiple-choice) enabled clear comparisons across all 150 public tenders.
Transparency and Traceability
The inclusion of exact text references enabled quick verification and deeper analysis. The Excel deliverable provided clear, standardized data for stakeholder review and research integration.
Scalability for Future Projects
The approach (Python automation + LLM prompts) can be adapted to other large-scale document analyses, positioning the organization for similar tasks without starting from scratch each time.