As a leading health insurance marketplace, GoHealth is dedicated to creating best-in-class technology solutions that help licensed insurance agents work more efficiently and effectively on behalf of consumers. One of those solutions is PlanGPT, a proprietary tool that leverages a large language model to look under the hoods of multiple Medicare Advantage plans. PlanGPT makes it quick and simple for agents to find the right answers to Medicare consumers’ questions about their coverage options.
In an increasingly complex Medicare market, PlanGPT simplifies shopping and enrollment. Some of the millions of people who get in touch with GoHealth every year are interested in switching from Original Medicare (the public insurance program for U.S. adults who are over 65 or have certain disabilities) to a Medicare Advantage plan offered by a private insurance carrier. Others are already on a Medicare Advantage plan and looking for an alternative that better suits their needs.
GoHealth agents use PlanGPT to analyze the choices that are available through GoHealth’s marketplace and make recommendations that fit a caller’s unique situation.
“At GoHealth, we are constantly innovating and refining our processes so our agents can provide the highest caliber of service,” GoHealth Chief Operating Officer Michael Hargis said. “PlanGPT empowers agents to zero in on the plan features that make a real impact in people’s lives.”
Choosing the Ideal Plan in a Complex Market
Different Medicare Advantage plans are available in every ZIP code across the U.S. The average Medicare-eligible adult had access to 42 plans in 2025 with variations in costs, coverage, and benefits that can be difficult to understand and compare.
Over time, there are significant changes in these individual plans and in the federal rules set by the Centers for Medicare and Medicaid Services (CMS) to govern the entire Medicare market. During 2024 Annual Enrollment Period (from Oct. 15 to Dec. 7) several factors made it especially important for consumers to explore their coverage options
- After CMS announced a decrease in the payments to private insurance carriers, some decided to discontinue certain plans, affecting more than 1.8 million consumers.
- In other cases, carriers lowered their expenses by reducing benefits.
- As the federal government set rules to lower the overall costs for drugs covered by Medicare Part D and Medicare Advantage plans, many plans increased how much beneficiaries pay out of pocket when they pick up certain prescriptions.
- Consumers’ preferred doctors, hospitals, and pharmacies may have left their provider networks.
With so much change, it’s often challenging for Medicare consumers to set priorities and sort through their options. At a glance, it can even be hard to tell the differences among plans since the main source for details is a document called Evidence of Coverage, which is often more than 200 pages long and hard for the average person to understand.
GoHealth developed PlanGPT to extract information from the plans’ Evidence of Coverage document, allowing agents to address consumer-specific needs within seconds.
How PlanGPT Informs Agents and Consumers
PlanGPT employs a large language model (LLM) to bring GoHealth’s licensed insurance agents greater visibility into plan documentation. In turn, agents provide consumers with the guidance they need to make vital decisions about their healthcare.
An LLM is an artificial intelligence (AI) system that’s trained to understand and generate the kind of natural language that humans use to communicate all the time. GPT, which stands for generative pre-trained transformer, is a category of LLM that’s widely used to create text, images, videos, and data.
PlanGPT uses an LLM to extract the most relevant information from Evidence of Coverage documents. When a GoHealth licensed insurance agent types a single question, PlanGPT responds with key points from the documentation for each plan they are comparing. The agent then provides the consumer with personalized guidance in real time.
“PlanGPT is an intuitive and user-friendly way for agents to ask their plan-related questions and efficiently retrieve answers from the Evidence of Coverage document,” GoHealth Senior Product Manager Kristen Nemeth explained. “Agents can access nuanced plan details that are highly personalized to consumers.”
The advantages of PlanGPT include:
- Enabling agents to locate accurate information in moments.
- Bringing together summaries of the most relevant content for up to three plans at once, so agents can easily make comparisons rather than manually searching through several documents.
- Allowing even agents who are relatively new to the Medicare Advantage market to move quickly through the process of finding a consumer’s ideal plan.
Thomas Young, Director of Sales Enablement at GoHealth, found that PlanGPT demonstrated its value for the 160 agents on his team, many of them still early in their careers. During the busy 2024 Annual Enrollment Period, team members served consumers quickly while avoiding errors or miscommunication. Along with other tools and process improvements, Young estimated that PlanGPT contributed to shaving an average of 10 minutes off the length of a typical phone call from the previous year’s enrollment period.
“Having a tool that’s so easy to use was a huge benefit for my people,” Young said.
Innovating for a Better Plan Fit
During the 2024 Annual Enrollment Period, GoHealth agents connected with more than 2 million consumers and assisted more than 481,000 of them to enroll in a new plan. PlanGPT is one crucial component of GoHealth’s larger initiative to develop and deploy technology that successfully matches each of those consumers with the ideal coverage (even if that means just sticking with their current plan).
PlanGPT complements GoHealth’s PlanFit technology, which provides tailored recommendations for every consumer. Drawing on analytics from millions of interactions with Medicare consumers, PlanFit identifies the plans that are the closest match to an individual’s preferences in coverage, additional benefits, costs, and provider networks. Then, an agent can narrow down those options by using PlanGPT to delve deeper.
As Lead Machine Learning Engineer at GoHealth, Sanjay Sharma constantly strives to improve the speed and accuracy of these tools and ensure that they work together effectively.
“PlanFit is best suited for evaluating quantitative benefits, like premiums and copays, while PlanGPT provides deeper context on qualitative nuances, such as eligibility requirements and coverage restrictions,” Sharma said.
With the additional context provided by knowledgeable agents and PlanGPT, 62% of people who enrolled through GoHealth chose coverage that PlanFit ranked in their top three options.
PlanGPT provides agents with invaluable visibility into plan features and costs. By staying at the forefront of advances in AI and industry best practices, GoHealth equips agents to provide every consumer with exceptionally clear and accurate guidance right when they need it.
Sources
Medicare Advantage in 2024: Enrollment Update and Key Trends. KFF.
Changes in store for Medicare Advantage as open enrollment starts. CNN.
