How accurate is the lens?
Every evaluation is tracked. Every agree and disagree refines the model. This is the live calibration dashboard — the ground truth behind Blackbird Scope detection accuracy.
Accuracy is earned, not assumed.
The Mechanism
Every response sharpens the lens
Every time you agree or disagree with a score, you help calibrate the lens. That feedback loop is the foundation of ground truth.
The lens evaluates
Blackbird Scope runs its multi-pass analysis on an AI response — detecting M-codes, scoring F/P/R, and assessing risk patterns.
You respond
After reviewing the evaluation, you mark whether you agree or disagree with the lens assessment. That single action is a calibration signal.
The model refines
Agree/disagree ratios are tracked per flag, per status, per detection type. Over time, thresholds adjust. The lens gets sharper. Ground truth emerges from volume.