Insight
AI in Employee Benefits: From Chatbot Curiosity to the Most Important Decision Your Departing Employees Will Ever Make
I’ve been in the benefits space long enough to remember when “digital enrollment” was considered a bold move. Now I watch TPAs, carriers, and benefits platforms race to attach “AI-powered” to everything they do. Some of it is real. Some of it is marketing. But the trajectory is undeniable, and the use case I care most about, helping people navigate health coverage at the most stressful moment of their lives, is one where AI is finally doing something genuinely meaningful.
Let me walk through how we got here, where employers and employees actually stand today, and why the way When uses AI is different from what most people think of when they hear the term.
How AI Found Its Way Into Benefits
Not long ago, AI in employee benefits was mostly back-office stuff. Claims adjudication. Eligibility verification. Fraud detection. Useful, certainly, but invisible to the employee experience.
That changed fast. Rising healthcare costs (nearly four in ten employers are concerned about medical costs increasing beyond general inflation) and a workforce increasingly comfortable with digital tools created real demand for smarter benefits experiences. Insurers, TPAs, and platforms started building tools that could surface personalized recommendations, answer questions in plain language, and guide employees through decisions that used to require a phone call to HR.
AI and machine learning began being applied to benefits data to help companies understand the sources of costs and employee behaviors, leading to better interventions and cost management. What once required a consultant and a spreadsheet was starting to happen in real time.
Employees Are Ready for This. More Ready Than Most Organizations Realize.
Here’s something that surprises a lot of people: employees aren’t resisting AI-driven benefits tools. They’re hungry for better guidance. 85% of employees consistently don’t understand their benefit offerings. That’s not a rounding error. That’s the vast majority of your workforce making one of the most consequential financial decisions of the year while operating in the dark. More than a quarter say picking a plan makes them stressed, and another 13% say they aren’t confident they chose a plan that suits their needs.
Personalized benefits education was cited as the most successful measure employers have found for shifting employee mindsets to understand the value of their benefits. When AI delivers on that (actually answering the question in plain language instead of pointing someone to a 40-page plan document) the results are striking. Employees are three times more likely to enroll in ancillary and voluntary benefits when they have immediate decision support.
The broader adoption numbers back this up. AI use at work has nearly doubled in two years, with 27% of white-collar employees now reporting frequent use. Finance and insurance ranks among the highest-adoption sectors. Employees aren’t waiting for their employers to catch up. The question is whether their benefits experience is keeping pace.
Where Most AI in Benefits Falls Short
A lot of what’s being called “AI” in the benefits space is decision-support for active employees during open enrollment. It’s genuinely useful. Helping someone understand the difference between a HDHP and a PPO, or figure out if they’re likely to hit their deductible, is a real service.
But it stops at the moment someone leaves.
Employee exits, whether voluntary, involuntary, retirement, or reduction in force, represent the single most consequential healthcare coverage moment in the benefits lifecycle. And it is almost entirely unmanaged. The person who just lost their job gets a COBRA notice. That’s it. No guidance. No comparison. No advocate. Just a document telling them their previous employer’s plan is available at a premium that’s often two to three times what they were paying before.
The natural result is adverse selection. The people who elect COBRA are overwhelmingly the ones with ongoing health needs, exactly the population that drives disproportionate claims costs for self-funded employers and impacts loss ratios for carriers and stop-loss partners. AI could hasten this decision-making by analyzing employee health information and personal priorities to recommend a best-fit plan, but almost no one has applied that logic to the post-employment moment, where it matters most.
What When Does Differently
When is not an open enrollment tool. It’s not a chatbot for active employees. And it’s not a COBRA administrator. That part is important, because the first question every TPA and COBRA admin asks us is whether we’re trying to replace them. We’re not. We extend what they do.
When sits at the moment an employee separates. Our AI-powered platform, built around Jamie, our plan navigator, guides departing employees through a personalized comparison of their real coverage alternatives: ACA marketplace plans, private options, Medicare for those who qualify, and ICHRA where applicable. The goal isn’t to push someone toward any particular outcome. It’s to make sure they have the information to make a genuinely informed decision about something most people have never had to think about before.
That’s exactly how we think about it. Jamie gets the conversation started, available 24/7, trained to explain plan differences in plain language, able to surface options someone would never have found on their own. But our licensed broker concierge team handles the moments that require human judgment and real advocacy. The AI doesn’t replace that relationship. It makes it better.
The outcome matters. Employers whose departing employees find better-fit, lower-cost coverage outside of COBRA see up to an 80% reduction in COBRA-related costs. For self-funded employers, that’s a direct dollar-for-dollar impact on plan performance. For stop-loss carriers and TPAs managing self-insured clients, it’s the difference between a clean renewal and a difficult conversation about why that one claim wiped out the year.
The Right Question to Be Asking
Most of the AI conversation in employee benefits is still focused on making open enrollment better. That’s a worthy problem. But there’s a far larger, far less managed opportunity sitting right at the end of the employee lifecycle.
Over 60 million employees transition out of the workforce every year in the United States. The vast majority navigate that moment alone, with inadequate information, making a coverage decision that could cost them and their former employer significantly more than it needs to.
AI should be doing something about that. That’s what When was built to do.
