Enter password to view case study
Product Design
/
b2b
/
AI
Aurora
Building trust in a AI-native platform across a multi-stakeholder ecosystem
Overview
As the sole product designer and first design hire, I led product design from 0 -> 1 at Aurora, building a multi-sided AI native platform for multiple stakeholders. I owned UX strategy, ran majority, if not all user interviews, built the design system from scratch, and shipped every screen.
The platform replaces a manual underwriting chain (email, re-keying, spreadsheets, weeks of delay) with an AI-powered pipeline: intake, extraction, enrichment, underwriting, quoting, binding. The hard part was not the automation. It was designing one system that serves three stakeholders with fundamentally different needs from the same data.
Year
2024 – Current
Company
Aurora
Role
Lead Product Designer
Scope
0 - 1
UX Strategy & Research
Design System creation,
Interaction design
Prototyping
What was the impact?
fundraising
The problem
Every day a decision sits unfinished is revenue that walks to a competitor.
Unstructured data everywhere
Data arrives as PDFs, spreadsheets, emails. Every handoff introduces friction, errors, and delay. No single source of truth.
No shared system
Every actor uses their own tools. Data is re-keyed at every handoff. One case might be typed into 3–4 different systems before it’s quoted.
Email as queue
The inbox is the work management system. Cases get lost in threads. No priority, no status, no SLA tracking.
Low transparency
Decisions live in emails. No structured record of why a case was priced, why a referral was approved, or what data the decision was based on. Compliance risk.
The target audience
Two users, one system
Each stakeholder has different goals, different risk tolerances, and a different definition of success.
The broker (intemediary)
wants speed and low friction. Submit in under five minutes. Quote back fast. No new tools.
What success looks like: Submit in under 5 minutes. Quote back fast.
The underwriter (expert operator)
wants structured data, fewer manual tasks, and clear rationale for every AI-assisted decision.
What success looks like: All evidence assembled. Review & price fast. Decision is defensible.
the pivot
What we found
From our research we heard a lot of top frustrations like “Too many portals to remember.” “Logging in takes longer than the submission.” “I already have the info in my email.” “Different portal for every insurer.”
The features converted. The portal didn't. The access pattern was the real barrier to why we couldn't drive more submissions.
AI enriched email reply
As a broker, I want to submit cases without logging into another portal.
A broker sends an email as they always have. The AI reads the attachments, extracts structured data, identifies gaps, and sends back a smart reply.
ux principles
Four principles to make fast decisions across the platform
recognition
over recall
Prefer suggestions, pre-filled answers, and templates over blank forms.
progressive
disclosure
Show the minimum to move forward. Reveal depth only when the user asks.
trust
calibration
For high-stakes steps, always show what we recommend, why, and what evidence supports it.
Automation without ambiguity
Auto-fill only when confidence is high. Otherwise, suggest and require confirmation.