You cannot price what nobody can prove.
Since January 2026 standard business policies exclude AI, so every renewal moves a company deploying agents from quietly covered to explicitly uncovered. Carriers cannot close that gap either: agentic liability has real demand, no loss history and no controls anyone can verify. Aegis is the part that makes the risk visible enough to write.
Agentic liability is where cyber was two decades ago. Cyber became $16.6B in global premium once the market could see inside the risk (2024, Guy Carpenter estimate). Aegis is the seeing-inside part.
Controls you can warrant
A checkpoint that provably ran, with signed rule versions and enforced fail modes. Runtime records, per action.
Warranty-grade, at bindClaims evidence both sides trust
Chain heads are countersigned by the insured and by Aegis, so a claim starts from a shared record instead of a dispute about one.
Countersigned, at claimThe loss-history substitute
Categorical telemetry from day one: agent type, action class, verdict, jurisdiction, industry. Zero PII, zero content.
Portfolio view, in betweenA policy is written once. The agent changes every week.
You bind on an attestation and assume it holds for a year. Aegis reads the policy itself, turns it into rules, and checks every agent action against them while the cover is live.
The insurer shall not be liable for any claim arising out of automated communications sent without a record of consent.
The insured shall give written notice of any circumstance within 72 hours of discovery.
Coverage territory: the United States and Canada.
A sublimit of $250,000 applies to regulatory proceedings.
- consent.recordrequired before any automated message, or the loss falls outside cover
- notice.window72 hours from discovery of a circumstance
- territoryoutside US and CA, the action is uncovered
- sublimit.regulatory$250,000, escalate before exposure passes it
Illustrative wording. Every pack is compiled from the insured's own documents, and never pooled or shared.
A record that the check ran on every action, not a questionnaire answer.
When an agent starts doing something you did not price, you hear about it while the policy is live.
Countersigned by the insured and by Aegis, so the claim starts from shared facts.
Priced on how this agent actually behaved, not on what similar companies tend to do.
Motor insurance did this with telematics. Aegis is the same idea for agents, reporting on lawfulness rather than speed.
Rails, not a carrier.
Aegis does not sell policies and never will. We make agentic books writable for the people who do: verifiable controls at bind, countersigned evidence at claim, portfolio telemetry in between. Priced with confidence instead of priced with padding.