Venu Mathukumalli · 813.600.8927 · Venu.Mathukumalli@gmail.com · linkedin.com/in/venumathukumalli
Case study 01 Quick peek 2 min read
B2B2C client app Single sign-on for client companies and their employees
An AI workflow for claims: the same seven steps for every claim. The AI handles the routine volume, and a licensed adjuster approves every payout.
The problem
Insurance carriers are moving claims triage toward AI-assisted decisioning — but a wrong payout or missed fraud signal creates real liability. The design question: let AI handle routine volume while keeping a human demonstrably, auditably in control of every dollar that moves.
The approach
Grounded in documented human-in-the-loop patterns — propose → review → commit, confidence-based gating, and escalation with full context handoff — applied to insurance-specific constraints: mandatory human approval, inspectable reasoning, and a permanent audit record.
The centerpiece screen
Before any action executes, the adjuster sees the full AI-proposed decision — payout, confidence, fraud risk, and the reasoning behind the number. Nothing is approved automatically, and nothing is approved blind.
Outcomes, business impact & learnings
What I delivered
Designed, not measured
What I took from it
A scoped concept with a clickable prototype, not yet tested with adjusters. It is positioned honestly as growth-area depth in AI workflows and human-in-the-loop UX, built on 9 years in regulated insurance platforms at Northstar.
Want the full walkthrough? All six screens, the design principles, source citations and a clickable prototype are in the deep dive on my portfolio site.
All six screens, the design principles applied, the interactive prototype, and full source citations.