BillShield AI

• 

  Front End Page of BillShield AI

Inspiration

Bills are often confusing, stressful, and full of vague charges. People may see terms like facility fee, administrative fee, late fee, out-of-network service, or roaming charge without knowing what they mean or what to ask before paying. We built BillShield AI to turn confusing bill text into clear next steps.

What it does

BillShield AI lets users paste bill text from a medical bill, utility bill, phone bill, insurance bill, or service invoice. The app analyzes the bill and returns:

• Bill type
• Provider or company
• Amount due and due date
• Risk score and risk level -Cross Checks Standard Costs
• Plain-English summary
• Charges to verify
• Questions to ask billing support
• Draft clarification email

Instead of blindly paying or ignoring a confusing bill, users get an actionable report they can use immediately.

How we built it

We built the frontend with Lovable and connected it to a Jac backend. The Jac backend exposes an analyze_bill function that receives pasted bill text, sends it to Featherless through an OpenAI-compatible chat completion request, normalizes the model output into structured JSON, and returns the result to the dashboard.

The final architecture is:

User pastes bill text
        ↓
Lovable frontend
        ↓
Jac backend: /function/analyze_bill
        ↓
Featherless AI model
        ↓
Structured JSON response
        ↓
BillShield AI dashboard

Built With

• chat
• completions
• featherless
• jac
• jaseci
• lovable
• openai-compatible
• python
• react
• tailwind-css
• tanstack-router
• typescript