Linkedin-in

CASE STUDY

Designing an Agentic Insurance Platform

Agentic workflow orchestration for modern policy administration systems.

Traditional underwriting workflows rely on rigid, step-by-step processes that require manual intervention at nearly every stage. This concept reimagines submission processing as an adaptive system where AI continuously analyzes incoming data, evaluates confidence, and advances the workflow automatically when possible while seamlessly involving underwriters when judgment is required.

Explore the System

Applying the Model to Submission Workflows

To validate the orchestration model, I applied it directly to the beginning of a submission workflow.

Instead of a traditional multi-step form, the experience becomes a guided process where AI continuously evaluates completeness, eligibility, and confidence — while the user contributes incrementally.

  • AI interprets partial input to establish early understanding
  • Copilot guides next steps based on confidence gaps
  • Workflow progression unlocks as confidence thresholds are met


This shifts submission from static data entry to a dynamic, intelligence-assisted workflow.

View Interactive Prototype

The Problem

Traditional underwriting workflows are rigid, sequential, and require excessive manual data entry.

Automation exists, but it is typically rule-based and brittle.

As AI capabilities mature, underwriting workflows should evolve from linear processes to adaptive systems.

Opportunity

Reimagine submission processing as an adaptive underwriting system.

AI continuously interprets submission data, evaluates completion confidence, and determines when the workflow can safely advance.

  • Continuous evaluation of submission completeness
  • Automated workflow progression when confidence thresholds are met
  • Human intervention only when expertise is required
  • Automation resumes seamlessly after human updates


The goal is to increase underwriting velocity while maintaining transparency, trust, and control.

System Architecture

The platform introduces an orchestration model where AI continuously evaluates submission progress and determines when the workflow can safely advance.

AI orchestration layer

  • Interprets submission inputs and unstructured data
  • Evaluates automation confidence across workflow steps
  • Populates and progresses workflow stages when thresholds are met


Human collaboration layer

  • Underwriters intervene when confidence drops below thresholds
  • Subjectivities and expert decisions are introduced
  • Approvals and adjustments refine the submission


Confidence loop

  • AI continuously reassesses submission completeness
  • The workflow advances automatically when confidence thresholds are satisfied

Platform Outcomes

Design principles and architectural decisions that enable AI-assisted underwriting workflows.

Experience Principles

Human-Centered AI Collaboration

  • AI operates continuously in the background
  • Users collaborate with AI rather than triggering automation
  • Humans retain control over critical risk decisions
Workflow Automation

Confidence-Driven Process Advancement

  • AI evaluates submission completeness and confidence
  • Automation progresses workflows when thresholds are met
  • Underwriters intervene only when expertise is required
Operational Benefits

Reduced Manual Workflow Overhead

  • Automation removes repetitive administrative tasks
  • AI assists with data interpretation and workflow progression
  • Underwriters focus on evaluating risk rather than assembling data
Platform Impact

AI-Orchestrated Underwriting Operations

  • Faster underwriting cycles and improved submission throughput
  • Scalable automation across underwriting workflows
  • Improved decision support for complex risk evaluation

Project Overview

Role

Lead Product Designer (Concept & Experience Architecture)

Scope

AI workflow architecture, submission orchestration model, underwriting experience strategy, and platform design alignment

Team

Design lead managing 2 direct reports. Collaborated with Product and Engineering to explore AI-driven submission processing and underwriting workflow automation

Timeline

Concept exploration → Fall 2026 product roadmap discussion

Next Case Study