Ashleigh still remembers the kind of calls that come when benefits confusion turns to panic.
An employee whose baby was just born and didn’t know if the newborn was covered. Someone who realized they missed open enrollment because they didn’t understand what “active enrollment” meant. Employees paying for benefits they didn’t actually understand, sometimes for years.
These weren’t rare situations. They became routine.
Today, Ashleigh is the Product Manager behind Flimp’s Employee Benefits AI Agent, building AI designed to answer these kinds of questions.
There’s a big difference between an AI system that works in a product demo and one that holds up when an employee needs a clear, trustworthy answer about their benefits. That difference can come down to one simple thing: whether the person building it has actually lived the problems HR teams face every day.
She Didn’t Start In Tech
Ashleigh’s path to building AI started somewhere most product managers’ stories don’t: a classroom.
“I originally intended to become a teacher and even pursued graduate studies in education,” she says. “But when I realized that wasn’t the right path, I pivoted into employee benefits through a connection.”
That pivot landed her at a small brokerage. Because of its size, she quickly found herself working across almost every function in the building. Account management. Compliance. Carrier coordination. Client communications. She didn’t silo herself into one lane. She learned benefits administration by doing all of it, often at the same time.
“I had the unique opportunity to wear multiple hats. I learned the ins and outs of benefits administration from the ground up.”
That foundation eventually led her to an in-house HR role managing benefits for more than 6,000 employees at a large public school district. That jump in scale brought a new set of lessons with it.
What 6,000 Employees Taught Her
Managing benefits at that scale has a way of surfacing truths that smaller environments can hide.
The questions Ashleigh fielded weren’t the ones she expected. Employees weren’t asking how to optimize their coverage.
They were asking what a deductible is.
Whether they were looking at the right document.
When they needed to enroll, they sometimes did so well after the window had closed.
“What surprised me most was how often we’d receive incredibly fundamental questions across all demographics,” she explains. “It didn’t matter if the person was early in their career or nearing retirement, highly compensated or hourly wage—there was a consistent, widespread lack of understanding when it came to benefits.”
The moment that stuck with her most wasn’t a complicated compliance question. It was simpler than that. Employees would call asking whether they had critical illness coverage. When told yes, they had no idea what it was or how it worked. They’d been enrolled in the benefit, paying for it, with no real understanding of what they had.
“What that revealed to me is that no one is ever really taught how benefits work,” Ashleigh says. “We assume employees know this stuff because it’s part of adult life. But the truth is, most people are navigating it blindly.”
This wasn’t a reflection of employee intelligence. It was a sign of failed communication. And it played out quietly across every industry, every organization, every year. Employees were enrolling because it was time, not because they understood their options. They were choosing plans because something looked affordable, not because they knew how it would actually function when they needed it.
For Ashleigh, this became the problem to solve. The Q&A volume and the HR team’s workload were obvious needs, but the deeper disconnect was between the benefits offered and what employees actually understood about them. Year-round educational tools, like benefits microsites and plain-language guides, can close some of that gap. But she kept coming back to one idea: what employees really needed was a place to ask their questions without fear of judgment and get an answer they could trust.
From Insight to Product Philosophy
When Ashleigh joined Flimp and eventually took the lead on developing the Employee Benefits AI Agent, she brought all of that with her.
Not as abstract lessons, but as lived scenarios, she could test the product against. Throughout development, she ran the product through hundreds of real-world situations she’d encountered as a broker and HR professional. They were edge cases, compliance gray areas, and the kinds of questions that seem simple until you realize how easily the answer can go wrong.
“I wasn’t just approaching this as a product manager,” she says. “I was testing and pressure-checking the chatbot against real-life scenarios I had lived.”
The process shaped the product’s three core principles: clarity, compliance, and trust.
Clarity Came First
Employees who are overwhelmed or embarrassed don’t need more jargon. They need plain-language, conversational responses and room to ask follow-up questions without feeling like they should already know the answer. The chatbot was built to speak the way a good HR professional speaks and not the way a policy document reads.
Compliance Was Non-Negotiable
Anyone who has worked in benefits knows the line you’re not allowed to cross. You can inform, but you can’t advise. You can explain what a plan covers, but you can’t tell anyone which plan to choose. That distinction matters legally, and it matters for employee trust. Developers built the AI Agent to stay on the right side of that line. The system ties every response to a specific source document, includes legal disclaimers, and informs without speculating or recommending.
“I considered not just the accuracy of the information, but how it was presented,” Ashleigh explains. “Could it be misinterpreted as a recommendation? Was the documentation it linked to current and correct?”
Transparency Builds Trust
Every answer the chatbot delivers includes the source it pulled from, so employees aren’t just taking the system’s word for it. They can instead verify the source themselves. In high-stakes moments, that sourcing instills confidence in the employee and their decision.
Why a Closed System Matters
One of the most consequential design decisions Ashleigh pushed for was keeping the AI Agent a closed system. That means it’s trained exclusively on an employer’s approved benefits content, not on general internet data or broad HR knowledge bases.
The distinction matters more than it might seem. An open AI system like ChatGPT can produce confident, well-structured answers that are completely wrong for a specific plan, a specific carrier, or a specific employee population. In most contexts, that’s a minor inconvenience. In benefits, it can mean an employee makes a coverage decision based on information that doesn’t reflect their actual plan.
A closed system eliminates that risk. Answers are sourced directly from the employer’s own materials. Responses reflect actual plan rules, not generic definitions. And because HR teams have visibility into what’s being asked and can see where questions go unanswered, the tool stays accountable in a way open systems simply can’t.
“It mirrors how compliance-conscious professionals work,” Ashleigh says. “Informative, transparent, but never advisory.”
Experience Isn’t a Soft Advantage
It’s tempting to think of lived HR experience as a nice-to-have in product development. It’s a background detail that adds credibility to the pitch but doesn’t fundamentally change what gets built.
This is one case where that thinking breaks down.
The decisions that make an AI tool genuinely trustworthy, the guardrails around advice, the sourcing requirements, the tone calibration, and the anticipation of follow-up questions aren’t obvious from the outside. They’re learned. They come from standing in front of confused employees, from navigating compliance audits, from knowing what it costs when an answer is almost right.
“In this field, there’s simply no room for error,” Ashleigh says.
That’s not a marketing line. It’s the standard she worked under for years before she ever started building. And it’s the standard she brought with her when she did.