AI-Assisted Disability Claims Modernization
Adjudicators spent hours manually cross-referencing medical records, policy law, and thick manuals just to land on a single disability rating, then checked their work again, because one mistake meant a performance hit.
Duration
Jul-Dec, 2025
Client
Federal Contractor
Services
#
User Research
#
Interaction Design
#
AI/UX
#
Systems Thinking
#
Prototyping
#
Agile Delivery
Challenge
A high-volume federal claims process required workers to manually synthesize multiple complex sources to calculate disability percentages. The cognitive load was unsustainable and the margin for error was high.
Solution
Designed a human-in-the-loop AI feature that reads the medical documents, follows their manual's instructions, references policy, and surfaces a recommended disability rating, all with full decision reasoning and source citations visible before the adjudicator accepts anything. Progressed from concept through lo-fi validation into active development.
Impact
Made the AI's reasoning fully transparent rather than just surfacing an output. Adjudicators weren't going to trust a black-box score affecting claimants' livelihoods, and honestly, they shouldn't. Reframing the AI as a research assistant rather than a decision-maker was the path to genuine adoption.