Portfolio
Medicare InsuranceUSA
AI Plan Matching for Medicare Agents
A retrieval-grounded plan-matching system that turns five years of CMS plan data into an agent's knowledge base — so a licensed agent, live on a call, can find the right-fit plan for each client without guesswork or hallucinated details.
Right plan, live
best-fit options surfaced while the client is on the call
No hallucinations
every answer grounded in CMS source data via RAG
5 years
of CMS plan data in one searchable knowledge base
The challenge
- CMS releases new Medicare plans constantly, and licensed agents have to find the right-fit plan for each client across a sea of options — fast, while the client is on the phone.
- Doing that by hand is slow and easy to get wrong, and every agency wants the information presented its own way.
- Bring AI into the loop and a new risk appears: a hallucinated plan detail isn't just unhelpful, it's a compliance and trust problem.
What we built
- We build the agent a knowledge base from CMS plan data — not just the current year, but the last five years — so the full plan landscape is searchable in one place.
- Live on the call, the licensed agent enters the client's information and the system surfaces the right-fit plans for that client's needs.
- Every answer is grounded in the source plan data using retrieval-augmented generation (RAG), so the AI does not hallucinate plan details. The knowledge base stays the same across agencies; we customize how results are presented to each agency's workflow.
Retrieval-augmented generation (RAG)CMS plan knowledge baseAgent-assist copilotClient-needs matchingSource-grounded answersPer-agency presentation
Described by industry and region to protect client confidentiality. Outcomes vary by business process, data quality, and implementation scope. Full case details available under NDA.