BenefitPilot

Inspiration

Dental benefits are confusing even when someone already has insurance. Employees may know they have an annual maximum, deductible, and coverage percentages, but it is much harder to understand what those numbers actually mean when a dentist gives them multiple procedures with different costs and timelines.

We built BenefitPilot to turn that complexity into a clear decision. Instead of only showing coverage, BenefitPilot helps an employee understand how different treatment schedules affect estimated out-of-pocket cost, monthly affordability, and benefit usage across plan years.

What it does

BenefitPilot lets an employee upload a dental treatment plan PDF or enter treatment details manually. The application identifies the procedures and combines them with the employee's dental plan, current benefit usage, deductible, and network rules.

It then creates three treatment strategies:

Budget minimizes estimated out-of-pocket cost.
Balanced considers both affordability and treatment timing.
Fastest prioritizes completing eligible care as soon as possible.

The employee can compare estimated plan contribution, member cost, monthly payments, completion timing, remaining benefits, and next-year benefit usage.

BenefitPilot also provides a dashboard with projected benefit utilization and alerts employees when benefits are running low or when unused benefits may expire near the end of the year.

How we built it

We built BenefitPilot as a full-stack web application with a Python backend and a browser-based HTML, CSS, and JavaScript frontend.

The backend loads employee, plan, deductible, and benefit-usage data from SQLite seed data. Insurance calculations and treatment scheduling are deterministic so that the financial values shown to the employee remain explainable and reproducible.

For a procedure with cost (C), deductible contribution (D), and coverage rate (r), the estimated plan contribution is based on the covered portion of the procedure while respecting the employee's remaining annual benefit maximum.

We use Groq for AI-assisted treatment-plan extraction and natural-language explanations. The AI does not decide the insurance payment or choose the final schedule. Those decisions remain in the deterministic backend.

We also integrated provider lookup through NPPES, projected current-year and next-year benefit usage, and a shared alert system across the Dashboard, Treatments, and Alerts pages.

Challenges we faced

One of the hardest parts was separating AI tasks from financial logic. It would have been easy to ask an LLM to recommend a treatment plan directly, but that could produce inconsistent numbers. We instead used AI only where it was useful, such as understanding unstructured treatment text and explaining results, while keeping calculations deterministic.

Another challenge was optimizing across a benefit-year boundary. A treatment performed in January may use a new annual maximum, so we had to calculate current-year and next-year benefits separately while still showing the employee one understandable treatment strategy.

We also had to handle cases where Budget, Balanced, and Fastest could legitimately produce similar results. Rather than artificially changing the numbers, we made each strategy use a different objective and clearly explain when the same schedule satisfies multiple goals.

What we learned

We learned that benefit optimization is not just about finding the cheapest option. A useful system also has to consider timing, monthly affordability, network restrictions, remaining annual benefits, and how clearly the recommendation can be explained to the employee.

The biggest lesson was that AI works best here as an interface to complex information, while deterministic software should remain the source of truth for financial calculations.

What's next

With access to real employer and insurer systems, BenefitPilot could connect directly to live eligibility, claims, provider-network, and benefit data. The same approach could also expand beyond dental coverage into other employee benefits where people need help understanding how and when to use the coverage they already have.