Jul 18, 2026Jul 18, 2026 · 3 min readDay 56: Why evaluating the model is not enoughWhy evaluating agents means measuring the whole workflow users experience.3 min read
Jun 20, 2026Jun 20, 2026 · 3 min readDay 28: Why weak evidence should trigger recovery, not confidenceWhy weak evidence should trigger recovery, not a fluent answer with false confidence.3 min read
