Designing Human-Centered AI
Personalization, Trust, and Industry Recognition
The Situation
By late 2023, generative AI had officially taken over the conversation in insurance, but nearly every pitch boiled down to faster quotes and lower costs. It was all efficiency, all the time.
I saw a different opportunity: use AI to amplify what people do best.
At American Family Insurance, I challenged my team to explore how generative AI could make people feel more supported and in control of their insurance. One team member, Monica Gengler, focused on the personal insurance review (PIR), a moment full of potential and often under-leveraged.
A personal insurance review checks that a customer is properly protected as their life changes. Agents who offer them earn higher satisfaction and retention scores, even when the customer says no; the offer alone builds trust. But gathering what the review needs is manual and slow, so most agents can't do one for every customer, every year. In practice the reviews ran inconsistent, and too often got repurposed to explain a rate increase.
The question was bigger than speeding up the review. Could AI make people feel more connected and confident in their coverage, at scale?
“When generative AI showed up, it was like a magic wand. Why waste that power shaving minutes off workflows when we could use it to strengthen financial confidence at scale?”
— Ryann Foelker, GenAI in Insurance, Boston, 2024.
The Tension
Most insurers treated AI like a silver bullet for ballooning costs and sluggish cycle times. Their use cases came off the same list every vendor was selling, so the best any of them could do was keep pace.
Meanwhile customers were overwhelmed, agents were overworked, and far too many households were underinsured. Left alone, the industry was on track to make outdated experiences faster.
Everyone was racing on technology. I made the case for racing on design, and Gwen Olson, then Innovation AVP, backed it.
So I gave my team a seven-week Design School Challenge. Start with AI's capabilities, current and projected. Then design an original use case for the customer and the agent. The specialized capabilities came after, built in the proof of concept to serve the use case we designed.
Specialized tech capabilities built for original use cases lead the market, and design is where you start. From the talk Gwen Olson and I gave at GenAI in Insurance, Boston, 2024.
“So often when we see new technology, we look at it from a small lens. How can we apply it to this process or that process? Generative AI is the tool to get really big with.”
— Ryann Foelker, The AI in Business Podcast, Emerj, 2024.
The Method
At the Milwaukee Institute of Art & Design, my industrial design professor, Tom, had spent the sixties shaping how the plastics boom reached consumer goods. When the rotomolders' trade association gave him a commission, he turned it into our class project. Learn the material first: low tooling costs, wide freedom of shape and size, perfectly consistent wall thickness. Then find a real problem it could solve.
I was teaching swim lessons to pay for groceries, so I designed a hollow, water-filled platform that gave little kids a shallow place to stand and catch their breath between strokes.
My best friend was studying chemistry at the time, and on our weekly calls we kept finding that our classes lined up. She was learning to formulate materials with new properties for manufacturers. I was learning to take those materials and processes and make things people wanted. She made the material. I made the product.
When "use cases for generative AI" became the industry's favorite phrase in 2023, I thought of Tom. I redesigned his method around a new material, in a different environment, for another kind of engineering, and wrote it down step by step. That became the Design School Challenge.
The Work
Over seven weeks, every member of my seven-person team studied generative AI like a material, then designed a concept that used it to solve a big industry problem. They presented weekly for critique, and I demanded conviction: the moment someone stopped believing in their concept, they had to pivot.
My Design School Challenge produced seven original use cases for generative AI, each designed for American Family's own customers and agents.
Five principles, shown here, shaped the work.
-
We explored what generative AI could do, current and projected, before mapping customer pain points. That gave us inspiration and a realistic lens on feasibility at the same time.
-
We built for agents, to strengthen their role as trusted advisors in risk and protection.
Adoption was a critique prompt in week three, before feasibility and before the pitch: what does this concept give agents that makes them want to use it?
-
We aimed at financial confusion, missed coverage, and eroding trust. Every idea had to show the value it would create before it graduated to concept development.
-
Each concept was scoped for near-term testing and long-term scale. I handed the collection to our co-development partner under a fictional carrier name, so the concepts would be judged on value rather than on who made them.
-
Every concept had to create more than one kind of value: functional, emotional, and the kind of value that changes how a person sees themselves.
Because the value was intentional, we knew what to measure, so each concept shipped with its own measurement plan, using existing metrics where they fit and proposing new ones where the concept demanded them.
Two concepts moved to proof of concept. PIRfy was the first to move from there into development.
PIRfy (Personal Insurance Review for You) is a generative AI assistant that gets agents ready for coverage reviews: consolidated customer insights, a personalized script, and prompts that help agents and customers work together to calibrate coverage, delivered inside the CRM workflow agents already used. Monica Gengler designed it. We co-developed it with Deloitte, with input from American Family agents every week. The IP is American Family's, from concept to experience.
Measuring PIRfy where value lives. A typical CRM tracks records. It doesn't tell you whether a relationship is getting stronger. While PIRfy moved into development, one of my designers and I created a CRM metrics strategy that measures PIRfy's value on sentiment and effort alongside sales and retention. It tracks three relationships at once: how the company supports its agents and team members, how its products earn customers' trust, and how customers, through behavior and sentiment, see the company and the people who serve them.
The Results
Designed for Scale and Revenue Lift
PIRfy cut more than thirty minutes of prep from every review, a pace projected to more than double review volume. Double a conservative five reviews a week and each agent reaches 260 more customers a year. Across 2,400 agents, that's 624,000 households with the coverage they need. Agents in early testing called it "a grand slam for retention" and "the most excited I've been in a long time." One asked, "Can we have this tomorrow?" With adoption across the agency force, the projected value ran to multimillion-dollar annual impact, through retention, cross-sell, agent time returned to relationships, and tens of thousands more households getting a tailored review each year.
When I left American Family at the end of 2025, PIRfy was in active development.
Insurer Innovation Award – Americas Region
In 2024, PIRfy earned top honors from The Digital Insurer, beating finalists from AXA, Prudential, and RBC. Full pitch and award listing at The Digital Insurer.
Strategic Co-Development Partnership with Deloitte
The Design School Challenge portfolio led to a co-development agreement with Deloitte Digital. A Deloitte partner called PIRfy "the most exciting thing happening with AI in insurance."
Reuters Trailblazing Women in Insurance
In 2025, Reuters Events named me one of its Trailblazing Women in Insurance on a nomination built around my Design School Challenge.
Why It Worked
We started with the material, not the industry’s use case list.
We learned what generative AI could do, then designed original use cases for the people who'd have to want it, then measured the results where the value lived. The product won an award. The method developed the team.
“Our job isn't to be everywhere. It's to show up spectacularly in the moments where support matters.”
— Ryann Foelker, Digital Insurance Connect, 2025.
This case study reflects work I was proud to contribute to at American Family Insurance. The content shared here is based on initiatives that were made public under executive direction and have been featured in industry presentations and publications since 2024, including The Digital Insurer’s World’s Digital Insurance Awards and the Reuters Events Smart Tech webinar. While I reference my work in innovation and design at AmFam, this content draws on publicly shared concepts, case studies, and my individual learnings, not corporate or trade-secret strategies. Recognition mentioned (e.g., industry awards or press coverage) is publicly available and is used here to illustrate broader themes in innovation and design leadership. All views expressed are my own. No confidential or proprietary information is disclosed, only publicly available outcomes, personal insights, and individual learnings