CYBER INSURANCE · TECHNICAL UNDERWRITING · QREATIVELAB

Price AI Risk
With Mathematical Proof

Every cyber policy on AI systems is priced on questionnaires. We give you formal proofs.

COBALT runs Z3 formal verification across 5 risk dimensions on your insured's codebase. Structured risk score 0–100. SAT/UNSAT proof per finding. Plugs into your pricing model via API. Per-assessment pricing — no subscription, no minimums.

Questionnaires tell you what they claim. COBALT shows you what the code does.

PARTNER WITH US →SEE THE SCORE MODEL
30+
Confirmed CVE disclosures
Z3
Formal proofs, not surveys
< 60s
Per assessment
API
Direct integration
MARKET CONTEXT

AI insurance is being written blind. That changes now.

📈
$17.6B
Cyber insurance market by 2028

AI-specific liability is the fastest-growing subcategory — and the least-priced segment.

🏛️
Aug 2026
EU AI Act enforcement begins

High-risk AI deployments require documented risk management. Underwriters need technical evidence.

🔍
0
Formal verification tools in underwriting today

Every cyber policy on AI systems is priced on questionnaires. No one uses mathematical proofs.

THE COBALT RISK SCORE

Five dimensions. One actionable number.

Each dimension is verified by Z3 SMT formal proofs — not pattern matching, not CVSS estimates. SAT means the vulnerability class is mathematically demonstrable in the target code.

D1
Integer & Arithmetic SafetyCWE-190/195
Overflow, wraparound, signed/unsigned cast errors in critical computation paths.
25%
weight
D2
Memory & Null SafetyCWE-476/131
Null pointer dereferences and buffer size miscalculations leading to crashes or heap corruption.
20%
weight
D3
Uninitialized StateCWE-457
Variables used before assignment on exception paths — silent failures in production.
20%
weight
D4
Cryptographic CorrectnessCWE-208/327
Non-constant-time comparisons, weak key derivation, improper algorithm selection.
20%
weight
D5
AI Agent Behavioral RiskE1–E11
Specification drift, authority escalation, hallucination under pressure — formal proofs via VERITAS.
15%
weight
SCORE → PREMIUM LOADING TABLE
80–100
LOW RISK
Standard market rate
60–79
MODERATE RISK
+15–25% loading
40–59
ELEVATED RISK
+40–70% loading or sub-limit
0–39
HIGH RISK
Decline or warranty required

Premium loading recommendations are indicative. Actual pricing decisions remain with the underwriter. COBALT provides the technical risk signal — you control the pricing model.

INTEGRATION

Plugs into your underwriting workflow in days, not months

01
API Integration

Your underwriting platform submits insured's repository or binary. REST API, JSON output. Under 60 seconds per assessment.

02
COBALT Risk Score

Z3 formal verification across 5 risk dimensions. Weighted score 0–100. SAT/UNSAT proof for every flagged path.

03
Premium Adjustment Signal

Structured JSON with score, band, dimension breakdown, and finding summary. Plugs directly into your pricing model.

POST /v1/assess → 200 OK
{
  "assessment_id": "COBALT-2026-UW-00147",
  "score": 58,
  "band": "ELEVATED_RISK",
  "premium_signal": "+40-70% loading or sub-limit",
  "dimensions": {
    "D1_arithmetic":   { "score": 45, "findings": 2, "proof": "SAT" },
    "D2_memory":       { "score": 70, "findings": 0, "proof": "UNSAT" },
    "D3_init_state":   { "score": 55, "findings": 1, "proof": "SAT" },
    "D4_crypto":       { "score": 80, "findings": 0, "proof": "UNSAT" },
    "D5_ai_behavioral":{ "score": 40, "findings": 3, "proof": "SAT" }
  },
  "scan_duration_ms": 34210,
  "data_retained": false,
  "chain_of_custody": "SHA-256: a1b2c3d4..."
}
DELIVERY OPTIONS

Works the way your team works

✓
REST API
JSON in / JSON out. Async webhook or sync response under 60s.
✓
PDF Report
Board-ready underwriting evidence package, chain of custody included.
✓
SARIF Export
Standard Static Analysis Results Interchange Format for your toolchain.
✓
White-label
Reports branded with your firm. Your analysts see your interface.
PARTNERSHIP PRICING

Per-assessment. No subscription.

You pay per policy assessment. Volume pricing starts at 50 assessments/month. White-label and API-only tiers available. No minimum commitment for pilot.

Pilot (1–10/mo)$500/assessment
Standard (11–100/mo)$350/assessment
Volume (100+/mo)Custom — contact us
White-label licenseAnnual flat fee
PARTNERSHIP INQUIRY

Become a COBALT Underwriting Partner

We respond within 4 business hours. Pilot program available — no commitment required.

Pilot program · No minimum commitment · QreativeLab Inc.