AI insurer Armilla AI ties its warranty to Lloyd's capacity, publishes no coverage limits

Armilla AI, an insurance vendor reflected in AI Dispatch's evidence-bound index, sells what it calls an "AI Performance Warranty" for generative AI and AI agent deployments. The company states on its site: "if an AI system fails to perform as promised, the warranty funds the customer remedy."

Armilla Insurance Services operates as "a Coverholder at Lloyd's," with the underlying "Affirmative AI Liability Insurance" underwritten by "certain underwriters at Lloyd's," according to the company. Separately, the company publishes a testimonial from a Swiss Re senior underwriter who says Swiss Re is "happy to support Armilla as a leading provider of accurate and thorough model verification" — an endorsement of Armilla's verification work, not a statement that Swiss Re underwrites the warranty itself.

The company frames its regulatory-violations coverage against two named statutes, advertising "Defense costs and fines for actions or investigations into AI underperformance, including 'high-risk' AI systems (e.g. Colorado AI Act, EU AI Act)."

Coverage is sold only "through properly licensed surplus lines brokers" and, the company states, "may not be available in all jurisdictions." One case study on the site describes an AI loan-decisioning deployment for banks and credit unions, MKIII, built on "an AI Loan Decision Model trained on millions of historical loans."

Armilla positions the product on two axes for corporate buyers: "Clear Compliance: Build confidence in your AI compliance with AI regulations and industry standards" and "Risk Transfer: Offload the financial burden of AI disruptions, from non-breach security issues to performance failures."

What the company's site does not publish, per the fields checked against it: no coverage limits, no liability cap, no indemnification terms, no stated dispute-resolution process for a warranty claim, and no audit-trail or tamper-evidence claim describing how a "failure to perform as promised" is verified and adjudicated. No compliance certification, third-party evaluation, or safety-evaluation methodology is disclosed on the page either.

For an index built to answer whether an AI deployment can be insured, audited, governed and paid on verified outcomes, Armilla answers the insurance half in detail — carrier relationship, warranty mechanism, named regulation — while leaving the mechanics of how a claim gets adjudicated undisclosed.