ibott — AV Risk Intelligence

GWP Monitored

$46M

+31% YoY

$0Target $75M

Active Operators

+4 this quarter

0Pipeline: 50

Safety Cases

2 pending

0Pipeline: 9

Portfolio Avg Score

0.27

0 best1.0 worst

Active Deployments

0Target: 50

Loss Ratio

41%

+14pts favourable

Target: 55%

AV Deployment Coverage

ibott Portfolio

Uber AVIP

AV Trucking

deriven.ai / RobotaxiTracker / NHTSA SGO

Loading deployment map…

Companies

Driverless

ibott GWP

Status

AV Companies

Waymo

9 cities

Aurora

7 routes

Zoox

2 cities

Motional

1 city

Nuro

1 city

Wayve

1 city

Kodiak

3 routes

Active States

California

Waymo, Zoox, Nuro

Arizona

Waymo

Texas

Waymo, Aurora, Nuro, Kodiak

Florida

Waymo

Georgia

Waymo

Nevada

Zoox

Pennsylvania

Aurora, Motional, Kodiak

New York

Motional

Comprehensive AV statute

Permit-based framework

Executive order only

No specific AV law

Waymo

Aurora

Kodiak

Zoox

Motional

Nuro

Wayve

Testing

Operator Intelligence

Waymo

Robotaxi

0.21

deriven.ai score

best in portfolio

Not AVIP

Live cities

RO Miles

>100M

Safety Case

1 ingested

Premium

$15M

Evidence coverage

Crash dataClaimsRegulatoryPre-data

Aurora

Autonomous Freight

0.31

deriven.ai score

improving

Not AVIP

Live routes

Corridor

TX / PA

Safety Case

1 ingested

Premium

$2M

Evidence coverage

GSN frameworkRegulatoryPre-data

Nuro

Last-Mile Delivery

0.44

deriven.ai score

limited data

Uber AVIPNo occupants

ODDs

1 active

Max speed

25 mph

Safety Case

1 ingested

Premium

<$1M

Evidence coverage

RegulatoryPre-data

Wayve

Robotaxi

deriven.ai score

pending

Not AVIPPre-regulatory

ODDs

0 live

Target

London 2026

Safety Case

1 parsing

Premium

Pipeline

Evidence coverage

UK AV Act filing in progress

Expanded view — Waymo

Waymo

SAE L4 Robotaxi — 9 live cities (US) / London (live) / 16 testing / Dubai pipeline

Robotaxi — SAE L4Not AVIPCA DMV verified

deriven.ai Score

0.21

+0.04 vs 2025 ▲

Crash Safety

84%

fewer injury crashes vs human

ODD Coverage

78%

22% in first-year cities, limited data

Data Quality

94%

aligns with 1 verified safety case

Underwriting Guidance

✓

Support renewal. 88% of 697 NHTSA incidents PD-only. BI hospitalisation 0.23/M mi vs 1.09 human. Zero passenger fatalities across >100M rider-only miles. Book performance within target.

↑

Defend against rate pressure. Only ibott operator with Swiss Re-audited, peer-reviewed retrospective claims data. 84% fewer injury crashes. Use this to hold pricing.

→

Price ODD expansion at renewal. 16 testing cities converting to live; early-stage markets include Atlanta, Miami. London and Dubai add new regulatory jurisdictions. Trigger premium adjustment clause per new ODD.

⚠

School bus SW — condition precedent for binding. 19 Austin incidents 2025-26, 5 post-fix. Require written fleet-wide SW rollout confirmation before renewal binds.

Maintain excess layer discipline. ADS frequency is improving. US nuclear verdict environment drives severity — do not compress excess on frequency gains alone.

ibott Risk Register

Passenger fatalities

0 to date

ADS engagement

Confirmed all incidents

School bus SW

19 incidents — fleet fix pending

Gates/chains recall

R25-001 resolved Nov 2024

London deployment

Live — 19 vehicles, TfL licensed

Dubai

Pre-filing — TBD

NHTSA reporting

Full narrative, not redacted

Performance vs Human BaselineDi Lillo et al. 2024, Traffic Injury Prevention

Fewer injury crashes

Waymo

Human

−84%

Airbag-deploy crashes

Waymo

Human

−73%

BI claims / M miles

Waymo

Human

0.08
vs 1.09

PD claims / M miles

Waymo

Human

0.23
vs 3.17

Methodology: Zip-code calibrated against human claims in Waymo operating areas. Swiss Re data. RAVE checklist compliant. Only ibott operator with independently audited, city-level retrospective claims data.

Incident Data (through 2025)NHTSA SGO

697

Total incidents

88%

PD only

Hospitalisations

Fatality

PD only

88% / 613

Minor injury

7% / 51

Hospitalisation

3% / 23

Fatality

<1% / 1

697 incidents reflect fleet exposure across ~100M miles, not fault attribution. Hospitalisations 0.23/M mi vs 1.09 human. Fatality was third-party, not a Waymo passenger. Full NHTSA narratives disclosed.

Safety Case Quality

Retrospective crash study — peer-reviewed, >100M miles, SF + Phoenix city-level statistically significantStrongest

Insurance claims benchmarking — Swiss Re zip-code calibrated, independently auditedVerified

NHTSA SGO full narrative disclosure — not redacted, supports underwriter due diligenceTransparent

2 recalls filed and resolved — SW version tracking active via NHTSA docketMonitored

School bus SW — second fix deployed Dec 2025, fleet-wide confirmation outstandingWatch

16 testing cities — no commercial permit, pre-data stagePre-data

ODD and Permit Coverage

California	CA DMV AV permit — commercial Verified
Arizona	AZ DOT permit — commercial Verified
Texas (4 cities)	TX DSHS permits — all verified Verified
Florida (2 cities)	FL DHSMV permits — verified Verified
Georgia	GA DOAS permit — early stage Monitoring
16 testing states	Manual testing — no rider permit Pre-data
London ↗	TfL operator licence — 19 vehicles live, 64 by end-2026 Live▸ ODD briefing
Dubai	Pre-filing — regulatory framework pending Pending

Waymo — London ODD Briefing

deriven.ai digital twin model · 19→64 vehicle fleet · 2025–2026

ODD Sub-score

0.29 +0.08 vs portfolio ▲

Worst-case multiplier

1.75×

evening + rain vs expected

Scenario volatility (CV)

0.41

threshold for stability: 0.15

CVaR (worst 25%)

1.57×

expected cost

Geographic HHI

0.04

mild concentration

Key Hazard Drivers — London ODD

Pedestrian density

19%

Pedestrian crossing type

15%

Traffic level

14%

Bus stop context

12%

Signal count / road efficiency

Bike lane presence

Underwriting Guidance — London ODD

⚠

Excess layer — apply 1.75× loading for evening/rain exposure. CV of 0.41 is well above the 0.15 stability threshold. A single point estimate is insufficient; widen margin accordingly.

→

Scale premium linearly with mileage. Cost elasticity is exactly 1.0 across 2025–2026 — no structural risk shift as fleet grows 19→64 vehicles. Rate per km is stable.

✓

Geographic accumulation is manageable. Top 5 postcodes (E14, E16, E6, SE1, E1) carry 37% of cost. HHI 0.04 — no single area dominant. Concentration stable year-on-year.

Monitor latent risk postcodes SE17 and N1C. Road hazard rank significantly exceeds current cost contribution. Fleet expansion into these areas will drive disproportionate adverse development.

Intersection Risk — Top 5 Archetypes (82% of total cost)

T-junction — 6 links, high ped. exposure35% cost / 12% mi

Large conflict zone · high crossing density · cost/mileage ratio: 2.9×

T-junction — uncontrolled priority, varying lanes22% cost / 14% mi

Uncontrolled priority · cost/mileage ratio: 1.6×

T-junction — uncontrolled, moderate ped.16% cost

Moderate pedestrian exposure · cost/mileage ratio: 1.5×

4-arm crossroads — signalled, high ped.5% cost

Signal controlled · cost/mileage ratio: <1.5×

Yield T-junction — residential4% cost

Low complexity · cost/mileage ratio: <1.0×

Geographic Concentration

0.04

HHI — no single postcode dominates

0.62

Gini — few postcodes carry most cost

37%

Cost in top 5 postcodes

30/86

Postcodes to cover 80% cost

Top 5 postcodes: E14, E16, E6, SE1, E1. Concentration stable 2025→2026 — fleet growth has not shifted geographic distribution. Latent risk: SE17, N1C — hazard rank exceeds cost share; flag for pricing if fleet expands here.

deriven.ai digital twin model · TfL operator data · Score CI: 0.2937–0.2949

⚠ Latent risk: SE17, N1C

Auto-ingestion active — next NHTSA SGO report Q3 2026

Action: Confirm school bus SW fleet-wide rollout

deriven.ai AV Risk Quantification Platform

Confidential — prepared for Apollo ibott · May 2026