Research

Code review claims, with receipts.

Field reports, measurement notes, and practical review methods. Every comparative claim identifies its basis; every case study points back to code you can inspect.

MeasurementHow Veridical beats the leaderboard — and finds more real bugsThe measurement basis, same-PR comparisons, and defects behind our code-review benchmark work.August 30, 2026 · 11 min read

Case studyThe $100M defect frontier models missedA cross-repository firmware defect, why a fluent review missed it, and how reproducible evidence exposed it.August 29, 2026 · 9 min read

Field guideHow to evaluate AI code review toolsA reproducible scorecard for comparing real pull-request reviews instead of feature lists and demo prose.August 31, 2026 · 8 min read

Field guideHow to review AI-generated codeA practical workflow for checking contracts, state, dependencies, and runtime behavior before merge.August 31, 2026 · 7 min read