How we analyze candidates, score transparency, and protect non-partisan integrity. Every methodology decision is documented here — because you deserve to know exactly how our analysis works before trusting it.
Single-model analysis carries inherent biases. By using architecturally different AI systems trained on different data, systematic bias is dramatically reduced. Where both models agree, confidence is high. Where they disagree, we surface the disagreement rather than hiding it.
Both models receive identical instructions: "You are a nonpartisan political analyst explaining officials in plain, everyday language. Avoid jargon." No political leaning is encoded. The prompt asks for strengths, concerns, platform analysis, and an overall assessment — always balanced.
A third AI pass synthesizes both analyses, preserving agreements and explicitly noting disagreements. The consensus includes: Summary, Key Strengths, Key Concerns, Platform Analysis, Overall Assessment, and (for incumbents) a Trust Score comparing promises to record.
We welcome scrutiny. If you see a flaw in our approach or have suggestions for improvement, we want to hear from you.
Contact Our Team