What Can They Do Now That They Couldn't Before? Jay Bell on Turning AI Activity into Measurable Impact

1

From Parris Island to the Claims Desk

Jay's road into insurance did not start in insurance. He began at Ohio State, joined the Marine Corps, and served six years as an enlisted Marine from 1993 to 1999, starting at Parris Island. He was an assaultman, trained on a shoulder fired assault weapon built to take out bunkers and light vehicles. As he puts it, a lot can go wrong with shoulder fired weapons, and a lot can go absolutely right, depending on which end you are on.

After the Marines he finished his degree at Albertus Magnus College while working full time for a local telephone company, doing installation and repair. He was one of the first DSL installers in the country, then moved into business service tech work. He thought about going into management, but the union went on strike so many times in that period that he decided he needed to do something else. He was weighing law school against business school, and he lived in the Hartford area, where, in his words, you can throw a rock and hit an insurance person every five feet. Someone suggested claims.

So at 29 he started over as a workers' comp claims adjuster in The Hartford's local claims office, making about a third of what he earned in his last year at the phone company. James, for whom work comp is his baby in insurance, approved of the choice: complicated claims can last a lifetime, and it is not an easy way to enter the industry.

2

Trading Six Sigma for an Underwriting Education

Within about 18 months Jay was a team leader, a pattern that would repeat through his career: managing people who were more experienced than he was. He remembers the first time an adjuster with about 15 years of experience had to come to him to get reserves approved. It was, at best, uncomfortable.

He wanted to move to the side of the business where money comes in the door rather than going out. At the time, becoming an underwriter at The Hartford meant six months of training at one of three locations, and relocating was not an option for his family. So he made a trade. The claims side had just been through Six Sigma programs, he had studied the method in business school, and underwriting was about to start its own engagements. He offered to bring Six Sigma to the underwriting organization if it would train him on underwriting. That is exactly what happened.

Offered a few different roles, he chose construction, which he saw as the closest he could stay to a blue collar role within a white collar world. He has stayed close to construction ever since, with a stretch in programs from 2016 to 2022 that culminated in leading The Hartford's middle and large commercial programs.

3

Walk a Mile in Underwriting's Shoes

James admits he envies that resume. He has spent 25 years building software for insurance companies and wishes he had come up through the ranks of insurance first, because you need to understand the problem to solve it, and there is no better way to understand it than to do the work you are trying to automate. Jay agrees, and his operations team now acts on it. They have a strong preference, not a mandate, for bringing people with an underwriting background into operations.

"The operation serves underwriting, and it's incredibly difficult to understand how to serve them best if you've never walked a mile in those shoes."

Jay Bell

4

Same Playbook, Same Places, Same Results

Jay is speaking at ITC on "Claims Transformation Without Claims Chaos," and he starts with a disclaimer: he is not a technologist. He has never worked in IT. His path into AI was predicated on empowering the individual.

Looking around the industry, he saw everyone rolling out efficiency initiatives in the same places the industry has poured them into for the last 20 years: intake on the claims side and submission intake on the underwriting side. Now leaders are asking where the real return on investment is. His question back is why anyone expected a different result from the same effort aimed at the same spot, when the only new ingredient is the tool.

His belief is that a true and lasting advantage has nothing to do with the tech stack or vendor selection criteria.

That is why his focus with AI is driving AI literacy and helping individuals achieve greater impact and better outcomes. Technology, he says, is a tool that helps.

"It's all about empowering the people."

Jay Bell

5

Activity Is Not Impact

Think of every meeting with a lovely presentation on the wall and a chart going up and to the right. Someone announces that 2,000 people were trained. Jay's response is a simple question: what can those people do today that they could not do before? If you think about it ahead of time, he argues, it is not hard to set up the measurement, calibrate it and establish a baseline before you roll anything out. Most organizations simply skip that step.

With the tools available today, he says, that habit is now the problem.

"I've started to get more relentlessly focused on the difference between activity and impact."

Jay Bell

"We're just continuing to do the same things we've always been doing."

Jay Bell

6

Keep the Discipline, Hide the Method

James has spent years studying Lean and implementing it alongside technology. His favorite book on the subject is not The Toyota Way but 2 Second Lean, which argues that shaving two seconds off every small task adds up to real results. He would happily see a whole ITC curriculum on Lean, because there are a lot of shiny objects on the expo floor that do not solve real problems.

Jay loves Six Sigma, and he says it sits at the root of everything he does. But he learned early, working with underwriters around 2005 and 2006, that he had to uncouple the discipline from the DMAIC methodology and its tools and keep them behind the scenes. The more he leaned into the tooling with the wrong audience, the more underwriters pulled back. They have a natural aversion to anything that feels like robotic science, and they will quickly remind you that underwriting is an art and a science.

James adds the other risk: process improvement can become a bureaucracy of its own, where being a black belt turns into worshipping the process instead of the two core tenets, improving efficiency and eliminating waste. In most of the claims organizations he works with, he says, many problems can be solved without AI at all. But AI is a remarkable multi-tool, one that threatens to be either our best friend or the destroyer of all worlds.

7

No Mandate, No Money, and a Room Full of Citizen Builders

For Jay, the answer starts with the individual again, which means AI literacy. Zurich stood up a special interest group called DAILY, short for Driving AI Literacy. It launched without a mandate and without a budget, and it passed 1,400 members in about three months. Leaders of that group are now moving into more influential roles and helping shape how the company thinks about AI and its implementation.

His favorite example is Miles Wemken, a member of his team. Six to eight months before the recording, Miles was a casual AI user who mostly polished emails with it. Then the team faced a choice: run traditional time studies, or find a way to use AI to do them more effectively. Miles spent about 60 hours of his own time over a couple of weeks and built an agent in Copilot. It is now used across Zurich, including in some international markets, and Miles has become one of the company's top citizen builders. It all started with literacy, and with giving one person room for his own epiphany.

Zurich North America has rolled out Copilot's premium tier to all of its people, and a small group also has access to Copilot Cowork. James and Jay compare notes on how fast the frontier tools are moving, from Claude and ChatGPT to Gemini and NotebookLM, and agree that the capability gains of the last year have been remarkable.

"We're turning regular folks into real citizen builders."

Jay Bell

8

The Goat in the Machine

Then there is the book. The goat on the cover of The Goat in the Machine is Jellybean, and the one on the back is Peanut, both his. Jay has seven goats, three sheep, a livestock guardian dog, chickens and guinea fowl. The goats have cleared almost three acres of overgrown land on his property, to a very high degree of efficiency. And no, the goat does not stand for greatest of all time.

The book grew out of LinkedIn. Jay was posting ideas that went nowhere until a friend told him the ideas were good but nobody knew who he was, and that he should lead with his more unusual interests. He considered Brazilian jiu jitsu, which he has practiced for about 30 years, but it felt overplayed in business, and his style is more self deprecating. So he wrote about being humbled, repeatedly, by the animals he takes care of, and the lessons they share with leadership. People found it interesting. What started as a coffee table book for friends and family became a field guide to turning AI activity into measurable impact, built around his Enable, Elevate, Evolve approach.

The book includes a nine question assessment, Walk the Fence, that leaders can run with their teams. Jay recently facilitated it with a room of CIOs and CTOs at a GDS event, across four 90 minute sessions over two days, and the feedback was strong. Now he wants to hold himself to the standard he asks of others. The book is on Amazon, and he gives it away free at goatinthemachine.net. The price is feedback: run the assessment with your team and tell him what worked and what did not, so the next version gets better.

9

Key Takeaways

10

Connect with Jay Bell