Smart Renewals

Transforming AI-Powered Renewal Workflows at Scale

ThreeFlow is a benefits placement platform connecting carriers and brokers in one collaborative system

Impact at a Glance

  • First-to-market AI quoting tool for insurance carriers

  • 65% reduction in manual data entry

  • 92% compliance rate with ease criteria

  • $200 million in recurring premium facilitated since June 2024

  • Improved carrier retention and platform adoption

Role
Product Design Lead: led the end-to-end design process for Smart Renewals

Timeline: 6 months (2024)


Tools: Figma, Loom, Sheets, Miro, Jira


My Collaborators: Product, Engineering, AI, QA, Product Marketing, Sales, Support, Biz Ops

Overview

Smart Renewals reimagined how insurance carriers process benefits renewals, establishing the first AI-powered quoting experience in the industry. I led the end-to-end design of a solution that enabled users to submit quotes faster, more accurately, and with dramatically reduced manual effort.

This initiative fundamentally shifted Threeflow from a workflow tool to an AI-powered quoting platform and directly enabled the launch of Smart Proposals just six weeks later.

The Challenge

The challenge wasn't a broken UI—it was a fragmented quoting ecosystem with four deeply interconnected problems that required solving as a system rather than in isolation.

Problem 1: User Abandonment

Carriers were abandoning Threeflow due to excessive manual requirements that were no longer needed in their internal systems:

  • Platform abandonment in favor of email/phone workflows

  • Users didn't know where they were in the quoting process

  • Required information was hard to find, unclear, or incomplete

  • High drop-off rates and frustrated underwriters

Problem 2: Complexity and Variability of Employee Benefits

The employee benefits quoting ecosystem involves multiple interconnected variables that all influence each other. This complexity framework illustrates how user segments, carrier sophistication, document types, and quote complexity all interact to create the challenging environment Threeflow operates within.

Problem 3: Organizational Maturity

Threeflow lacked the sophistication and processes to deliver a user-centered solution:

  • Leader-led solution with light technical validation

  • Some carriers relied on internal team support; others had strong automation

  • Threeflow needed frameworks for cross-functional collaboration

  • We had to support both ends of the maturity spectrum

  • Required establishing new development processes for AI features

Problem 4: Technical and Data Debt

The existing system suffered from foundational issues that would undermine any AI improvements:

  • Out of date technical stack

  • Inconsistent design patterns across workflows

  • Busy UI that hid the work from the user

  • Brittle code in key user flows

  • Tech stack not prepared for AI

The Opportunity: Create an AI-powered experience that earns trust while reducing manual work, adapts to varying user and carrier maturity levels, and establishes a scalable foundation for quoting intelligence.

My Process

Research & Discovery

Research Approach

Rather than exploring open-ended solution options, this research focused on understanding how user pain points could be addressed within the constraints of our planned AI-powered approach.

  • Analyzed user behavior and 3 months of carrier feedback

  • Conducted interviews with carriers across a range of automation maturity

  • Audited ThreeFlow’s competitors quoting tools

  • Uncovered role-specific pain points and process breakdowns

User Variability

We uncovered significant variability in user preferences, quoting behavior, and trust in automation.

User Archetypes by Autonomy and Output Preference

“I would love help with quoting but I would still want to review before submitting my quote”

“Our internal system generates the proposal based and then I come to ThreeFlow I basically have to do the same thing but manually”

These archetypes confirmed that many carrier users would welcome help from AI — as long as the support respected their preferences for control, review, and throughput. These distinctions guided our Smart Renewal MVP design and helped us define how automation should flex across user types. The findings also pointed to future product needs based on quote sophistication, volume, and trust in automation.

Key Themes & Opportunity

Across interviews and data reviews, three pain patterns emerged:

  • Lack of clarity about where users were in the process and how data moved

  • Overloaded interfaces that overwhelmed users before they could understand the task

  • Duplicated effort from manual handoffs and redundant workflows

Together, these insights confirmed an opportunity to leverage AI to improve carrier workflows — not just to reduce effort, but to lay the groundwork for an intelligent quoting system — scalable, adaptive, and ready for AI.

Design & Strategy

I developed a multi-faceted approach that addressed UX, technical, and organizational challenges simultaneously:

User Experience Strategy:

  • Created video user journey maps and storyboards to build team empathy

  • Explored different UI strategies; with the team deciding to move forward with simplified the UI to reduce visual noise within technical constraints

  • Built prototypes to validate integration within real carrier workflows

Storytelling the Future State

Play the embedded Loom to experience the narrative shared with PMs, AI leads, and sales.

Storyboarding isn’t an output for me — it’s a key alignment method.

AI Integration Strategy:

To support a fragmented and high-stakes quoting process, I designed an AI strategy focused on clarity, trust, and human fallback. This strategy was grounded in user research and delivered in phases:

Phase 1 Surface Confidence + Status (Delivered):

Goal: Processing transparency and status clarity
We launched AI extraction and confidence surfacing alongside real-time status indicators, so users could understand where they were in the process and what the AI was doing.

Phase Two Enable Control + Clarity (In planning):

Goal: Interactive control and capability communication
We're designing lightweight interventions that give users control over quote accuracy thresholds, editing support, and fallback triggers — tailored to their quoting sophistication.

Human-Centered Safety Net

  • Threeflow Assist: I designed this as a human-in-the-loop fallback for quotes the AI couldn’t confidently support — ensuring reliability while reducing effort.

  • Ease Metrics: I partnered with product and data leads to define a new success lens for AI features — measuring user confidence, effort saved, and error reduction rather than system accuracy alone.

AI Rollout Plan

To implement the strategy incrementally, I led a cross-functional roadmap that aligned engineering, data science, UX, and operations across five delivery lanes. This phased plan prioritized visible impact and minimized downstream disruption.

The visual below maps the effort lanes, delivery scope, and associated carrier and BizOps impacts for each milestone.

Organizational Strategy:

  • Led cross-functional workshops for alignment across sales, support, and operations

  • Established Threeflow's first formal beta program

  • Created celebration systems to maintain momentum through complex development

Bridging Organizational Gaps: Early Workflow Model

One of the biggest obstacles to shipping Smart Renewals wasn't just technical or UX-related—it was organizational immaturity. Teams operated in silos, and no one had a shared view of how quoting really worked end-to-end.

To address this, I created an early behavioral workflow map that clarified:

  • Role responsibilities (Broker, Carrier, Threeflow)

  • Key logic gates (happy vs. unhappy paths)

  • Process breakdowns and redundancy points

  • Where automation might help—without assuming it was ready yet

This map helped engineering, CX, QA, and product leads:

  • Understand quoting logic before jumping to implementation

  • Collaborate using a shared artifact, not tribal knowledge

  • Identify where we lacked tooling, definitions, or feedback loops

Solution Exploration & Testing

Designing Smart Renewals meant navigating critical tradeoffs between automation and control, clarity and complexity.

I explored multiple approaches, including:

Exploration 1: Card-based

  • Leveraged existing spreadsheet-like patterns for fast onboarding

  • Familiar to users, but not scalable or clear for complex workflows

Exploration 2: AI and enhanced project management

  • Introduced quote task tracking based on user feedback

  • Could be helpful, but also could add more work to the quoting process

Exploration 3: Simplified workflow

  • Cleaned up the product UI and the experience to enhance the user’s feeling of simplicity and ease

  • Focused only on essential actions

Exploration 3 was the chosen experience to enhance the ease of the experience. The development team also chose the simplified workflow because it was quickest to implement and still have impact from a technical perspective.

Key Design Decisions

Designing for AI Gaps: Threeflow Assist

Since our AI models weren't ready to support every quote type at launch, I designed a fallback mechanism that became a core feature rather than a limitation.

The Solution: Quotes that couldn't be AI processed users were still able to route to our internal ThreeFlow Assist, BizOps team, to have someone generate the quote based on the uploaded information; thereby still providing quoting ease to the user.

Why This Worked:

  • Kept users in flow without blocking on AI limitations

  • Built user trust through clear handoff communication and resolution tracking

  • Collected real data to strengthen future model training

  • Enabled us to ship earlier and learn faster

This human-in-the-loop design became the model for AI-human collaboration across Threeflow's product suite.

Creating Flexible Workflows for Varied Product Maturity

Challenge: Threeflow processes quotes for different ancillary insurance products, and AI quoting wasn't able to support all product types at launch.

Solution: We introduced Threeflow Assist to handle unsupported products—allowing users to continue quoting even when AI couldn't help. This ensured partial automation benefits were still delivered at launch. Over time, we expanded AI coverage to additional product types, allowing the experience to scale with model maturity.

  • Ran the company’s first formal beta program

  • Delivered prototypes, embedded QA specs, and held async validation loops

  • Facilitated team alignment through vision walkthroughs and structured feedback rituals

The Solution

AI-Powered Document Processing

  • Automatic extraction of renewal data from varied document formats

  • Pre-filled forms with intelligent field mapping

  • 92% accuracy rate with clear confidence indicators

  • Seamless handling of document quality variations

Transparent AI Feedback System

  • Clear visual indicators showing AI confidence levels for each extracted field

  • Distinct treatments indicating information source (document vs. AI inference vs. user input)

  • Progressive disclosure: detailed processing information available when needed

  • Framework for handling both AI errors and ambiguous information

Streamlined Review & Submission Process

  • Simplified interface integrating into existing carrier workflows

  • Intelligent validation preventing common errors before submission

  • Role-based paths for underwriters and analysts

  • Adaptive submission supporting both manual review and automated processing

System-Level Architecture

The complete quoting process flows through automation and human support based on:

  • Carrier role and preferred quoting method

  • Document completeness and quality

  • AI confidence thresholds

  • Organizational maturity level

The Results

Quantitative Results

  • 65% reduction in manual effort

  • $200M+ in renewal premiums processed

  • 92% compliance with ease-of-use criteria

Qualitative

  • Increased carrier confidence in platform capabilities

  • Reduced platform abandonment in favor of email/phone workflows

  • Improved marketplace liquidity through increased participation

  • Enhanced Threeflow's competitive position as first-to-market AI solution

Business Value

  • Shifted Threeflow from workflow tool to AI-powered quoting platform

  • Established reusable design patterns for future AI features

  • Enabled Smart Proposals launching six weeks later using same architecture

  • Created scalable foundation for expanding AI capabilities to address Medical products as well as helping to automate onboarding with AI

Ripple Effect

Smart Renewals helped ThreeFlow become the first company to bring AI into the benefits placement space — and set new expectations for how intelligent systems should behave.

But more than a successful launch, this work created the foundation for system-wide advancement:

  • Sparked development of a company-wide design system

  • Gained leadership buy-in to invest in the Vue3 upgrade — unlocking improvements in front-end architecture and design craft

  • Demonstrated the strategic value of design in influencing technical roadmap decisions

  • Evolved AI from a prototype layer to a scalable, trusted infrastructure

What I Learned

This project demonstrated that successful AI features aren't about perfect automation—they're about making users feel confident and in control while dramatically reducing their workload.

Key Takeaways

  • Build fallback pathways early: AI needs support scaffolding to ship successfully

  • Trust through transparency: Users need to understand how AI makes decisions

  • User control is critical: Automation should enhance, not replace, user judgment

  • Frameworks scale: Well-designed patterns influence entire product strategies

  • Flexibility is essential: Enterprise products must adapt to varied organizational maturity

You don't need perfect automation to create value: Strategic human-in-the-loop design enables earlier shipping and faster learning

Smart Renewals wasn't just an AI feature—it was a designed system that adapted to user trust, organizational maturity, and automation readiness. By solving user experience, technical debt, and organizational challenges as interconnected problems, we created a solution that shipped fast, scaled thoughtfully, and established the foundation for Threeflow's AI-powered future.

The project's success came from recognizing that the best AI experiences feel magical not because they're fully automated, but because they seamlessly blend human intelligence with machine capability in ways that feel natural and trustworthy to users.