For judges · technical deep dive

3-tier hybrid verification

AI alone is not enough. Neither is human alone. Aqoryn combines deterministic checks, AI semantic verification, and staked human review — each tier catching what the others miss.

"AI cannot release funds. It only generates verifiable evidence."

⚙️
Tier 1

Deterministic checks

No AI. No subjectivity. Ungameable.

~$0 · < 2 seconds

If any check fails: confidence is penalised. Auto-release is blocked.

🤖
Tier 2

AI semantic verification

Completeness, not quality.

~$0.01/verification · 3–8 seconds

≥85%: auto-release. 75–85%: human review. <60%: dispute triggered.

AI cannot release funds. It only generates verifiable evidence.

👤
Tier 3

Human review (staked)

Quality. Judgment. Economic accountability.

Triggered when: AI confidence 75–85% OR job value ≥ $1,000

0.5–1% bounty paid to reviewer · up to 24 hours

Human verdict overrides AI for quality. Slashing prevents lazy reviews.

Routing logic

AI ≥ 85% + all deterministic pass Auto-release
AI 75–85%👤 Human review
Job value ≥ $1,000👤 Human review
AI < 60%⚖️ Dispute / DAO

Blending formula: final_confidence = (human × 0.6) + (AI × 0.4) when human review applies.

Judge Q&A