Use AI to clear a health-insurance claims backlog

    Easy20 minBreak it down · AI rubric, per criterionCFDE
    • clarify before building
    • success metric
    • scope control

    The brief

    Shrayan Health TPA (fictional) is a third-party administrator in Pune that processes reimbursement health claims for four insurers. You have been introduced to the CEO by your account manager.

    Facts shared so far:

    • 41,000 reimbursement claims are pending. Average turnaround is 19 days; the contractual SLA with insurers is 15 days, and two insurers have started applying penalties.
    • Claims arrive as scanned PDFs and phone photos: claim form, discharge summary, hospital bills, pharmacy bills, prescriptions, KYC. Mix of English, Hindi and Marathi, some handwritten.
    • 160 medical and claims officers review claims in a workflow tool built in-house in 2019.
    • Around one in four claims is sent back for missing documents, which restarts the clock.

    The CEO says: "Everyone is using GenAI now. I want you to use it to clear our backlog by March. What will you build?"

    You have 20 minutes with her. Write how you would break the problem down. The interviewer wants to see what you ask and how you define success before any solution.

    How it's graded

    • Asks clarifying questions before proposing a solution3 pts
    • Defines 'clear the backlog' as a measurable target with guardrails3 pts
    • Proposes a first step matched to the bottleneck, not a big platform2 pts
    • Covers risks: wrong payouts, health data, handwriting, adoption2 pts

    Your answer

    What would you ask first, and why? Rank them. Do not propose a solution yet.

    0/220 words

    State your working assumptions, then define the metric precisely: baseline, target, how it is measured, and guardrail metrics.

    0/180 words

    Only now: the options you would consider, tied to what the answers might reveal, and what you would build first.

    0/250 words

    What could go wrong, who could be harmed, and how you would contain it.

    0/150 words

    Graded by AI against the ADaSci rubric. Your answers stay private; only your score shows on your profile.