Uwazi: ML-Powered Information Extraction ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Lead UX/UI Designer | HURIDOCS | 24 months Team: CTO + 2 ML Specialists + 6 Engineers Challenge ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Human rights organizations were drowning in documentation but lacked the technical resources to leverage AI for systematic information extraction. Existing approaches forced organizations to either hire expensive technical specialists or manually review every document. Research & Insights ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ • Analysis of internal usage patterns and user feedback • Technical feasibility studies with ML engineers • Competitive analysis of existing AI document processing platforms • Workflow integration planning to enhance (not disrupt) current processes Key Features ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ • Workflow creation interfaces for building extraction models • Smart review flows for validating AI suggestions • Batch processing tools for managing large document sets • Structured data output integrated with user workflows Tools ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ MindNode, Figma, Miro, Document analysis tools Results ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ • Reduced data extraction time by 60% for non-technical users • Enabled organizations to train AI models without hiring technical specialists • Streamlined evidence-gathering timelines while maintaining accuracy standards • Adopted across Uwazi's user base of 4,000+ users in 150+ human rights organizations View full case study: juanmnl.com/projects/uwazi-ml