AI Assistant
docs: add quantitative evaluation benchmarks (80-question single-doc test) and report
b3242a4