ClauseHound: a pet-insurance decision engine that refuses to guess

ClauseHound is a pet-insurance decision engine designed to help consumers navigate the complex world of pet insurance policies. It takes the four most critical decisions pet owners make when selecting a policy and answers each one by citing the exact clause from the insurer's policy text.

When shopping for a pet insurance policy, users can ask ClauseHound whether they should get pet insurance or put the monthly premium money into a savings account. ClauseHound then runs the deterministic math (premiums vs. vet bills over time) and tells the user plainly when saving wins. It provides a lifetime cost comparison, showing the true 10-year cost per plan, including premiums, deductibles, and reimbursement math, exact to the cent.

ClauseHound also analyzes claim payments, answering questions like, "My dog needs a $4,200 cruciate surgery in month 8 — what would each insurer actually pay?" It answers carrier-by-carrier, citing the coverage clauses, exclusions, and waiting periods that govern that claim. Additionally, ClauseHound offers switching guidance, answering questions like, "My dog has a pre-existing skin condition; which insurers will still cover her if I switch?" It cites the pre-existing-condition clauses for each carrier.

The answer format is simple – verdict first, followed by the citations that support the decision. If the corpus doesn't cover the question, ClauseHound will respond with "I couldn't verify this in the policy documents" – a refusal, not a guess, with pointers to what is answerable. If coverage clauses and exclusions both match, both surface side by side, marked as unresolved. The agent never picks a winner for conflicts, leaving the decision up to the user.

ClauseHound's knowledge base consists of structured content in Sanity, modeled after the insurers' published US sample policies. The schema ensures that every clause document has a required sectionRef (e.g., Sec. 3.4) and a required policy reference back to its policy document, which references the insurer. This trust strategy guarantees that the agent can only access relevant content when answering questions.