Overview
Refreshed September 30, 2026.
Glacis Academy connects AI safety, runtime evidence, and insurance judgement. Learn how a consequential AI workflow can fail, what a control or signed record establishes, and how that information can help an underwriter, broker, or claims reviewer. The producer-licensing curriculum remains the foundation for regulated insurance activity. STABLE
Choose your route
Section titled “Choose your route”| Your task | Start here |
|---|---|
| Prepare for a market conversation | Lloyd’s market lab |
| Understand modern AI failure paths | AI safety in practice |
| Explain Glacis and OVERT | Inference boundary → OVERT |
| Learn underwriting and product mechanics | Underwriting fundamentals → Carrier economics |
| Study for a producer exam | Exam mental model → sections 01–12 and your state |
| Check a volatile claim | VERIFY hub |
Current context
Section titled “Current context”Lloyd’s announced Glacis Technologies as a member of Cohort 17. The academy uses that setting to teach market discovery and falsifiable pilots. Commercial adoption, capacity, wording, pricing, and coverholder authority are separate decisions. GROUNDED · Lloyds Official announcement, checked September 30, 2026.
The rule running through every route
Section titled “The rule running through every route”Control → execution → outcome → evidence. Ask what was intended, what ran, what effect was observed, and what can be independently checked. Never turn evidence of one link into an assertion about the whole chain. STABLE
A licence authorises particular regulated activities under applicable rules. Delegated underwriting authority requires an appropriate agreement and other required approvals. Selling software, helping investigate an incident, and underwriting insurance are distinct activities. STABLE Specific legal requirements remain VERIFY.
The earlier MGA strategy is background in section 15. The market curriculum keeps Glacis’s evidence role and the insurer’s risk authority distinct.