The Coverage Ontology

The semantic layer for the business of insurance—coverage as objects, relationships, and outcomes, not text.

Coverage is a living system. The model that understands it has to be one too.

One model of coverage. Everything ships from it.

The ontology is not a product you query. It is the asset Qumis is built on—the connected model of how coverage actually works, sitting beneath every agent, every report, every answer.

Isolated compute

No training on client data

Law

Market

Form

Record

Firm

Policy Endorsement Exclusion Sublimit Claim Contract Insured

The Qumis App

Certified Agents Reports Comparisons The Diagnostic

The coverage graph

PolicyEndorsementExclusionSublimitClaimContractInsured

The knowledge layers law · market · form · record · firm

Governance & privacy isolated compute · no training on client data

Every analysis draws on the graph. Every reviewed decision enriches it.

Coverage, as a coverage mind holds it.

The Policy Endorsement 04 Exclusion §B.1.g Sublimit Claim fact Contract §9.2 Grant §I.A Form CP 10 30 Additional insured Notice condition Case law Market benchmark

One ontology. A private instance per firm.

Firm A

The questionIs assault & battery excluded here?

Escalate—your playbook flags A&B sublimits on hospitality accounts.

Shaped by Firm A’s playbooks and panel standards

Firm B

The questionIs assault & battery excluded here?

Proceed—matches your manuscript endorsement standard.

Shaped by Firm B’s precedent and house style

Configured to the firm. Each instance carries your playbooks, panels, precedent, and house style—so the same question returns your judgment, not a generic one.

Compounding through the governed learning loop. Every reviewed decision makes the next one better—governed professional feedback, never training on your data.

General-purpose AI reads insurance. Qumis understands it—because understanding was built first.