Deal Flow

Traditional insurance is plagued by slow, manual claim verification and opaque rejection reasons. dealFlow solves this by acting as an AI Forensic Auditor.

Comment

Inspiration

🌊 DealFlow

AI-Powered Autonomous Insurance Protocol

Bridging the gap between on-chain liquidity and off-chain reality.


DEPLOYED LINK

https://deal-flows.netlify.app/

💡 Inspiration

Traditional insurance is plagued by slow, manual claim verification and opaque rejection reasons. dealFlow solves this by acting as an AI Forensic Auditor. It connects a user's real-world evidence (photos of bills/receipts) with a company's on-chain liquidity and policy rules, creating a system where valid claims are paid out instantly, and invalid ones are rejected with clear, transparent reasons.

📸 How It Works

1. The "Brain" (Automation Engine)

The core of dealFlow is a Make.com scenario that orchestrates the logic between Gmail, Google Gemini AI, and our Supabase ledger.

![https://eu1.make.com/public/shared-scenario/2CgSk0vkgBS/integration-gmail-http]image

2. The Verdict (AI Forensic Auditor)

![insurancebot45@gmail.com]image

🚀 Key Features

  • 🤖 AI Forensic Auditor: Uses computer vision (Google Gemini) to extract specific data from receipts (Merchant Name, Date, Total Amount) and cross-references it with the insurer's policy terms.
  • ⚡ Automated Claim Settlement: No human intervention required. If the data matches the policy, the claim is approved instantly.
  • 🏢 Dual-Dashboard System:
    • Business Dashboard: Real-time financial tracking (Total Locked, Paid Out, Available Balance).
    • Consumer Dashboard: Simple interface for users to track claim status.
  • 📧 Zero-Friction Submission: Users don't need to learn complex dApps to file a claim; they simply reply to an email with their evidence.

⚙️ How It Works (The Flow)

  1. Business Onboarding: An Insurance Agent registers on the Business Dashboard and funds their liquidity pool.
  2. Claim Submission: A user submits a claim via the Consumer Portal or by sending an email with an attachment (photo of the bill).
  3. The "Watcher": Our Make.com engine detects the new submission.
  4. AI Analysis: Gemini AI scans the image, acting as a forensic auditor. It extracts the Bill Amount and Wallet Address and checks against the specific Company Policy stored in Supabase.
  5. The Verdict:
    • Approved: The system updates the on-chain ledger and deducts the amount from the company's "Available Balance."
    • Rejected: The AI drafts a specific reply explaining exactly what is missing.
  6. Notification: The user receives an instant email notification with the verdict.

🛠️ Built With

📦 Getting Started

Prerequisites

  • Node.js (v16 or higher)
  • npm or yarn

Installation

  1. Clone the repo sh git clone [https://github.com/krish2413179-prog/dealFlow.git](https://github.com/krish2413179-prog/dealFlow.git)
  2. Install NPM packages sh npm install
  3. Start the development server sh npm run dev

🔮 What's Next for dealFlow

  • Smart Contract Integration: Moving the ledger from Supabase to a fully decentralized Smart Contract on Polygon/Ethereum.
  • Fraud Detection 2.0: Implementing advanced metadata analysis to detect Photoshop-edited receipts.
  • Multi-Chain Support: Allowing payouts in USDC, ETH, or MATIC.

👥 Team

  • [Krish Sharma] - Full Stack & Automation

Note to Judges: The Make.com scenario JSON is included in the /automation folder for review.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Untitled

Built With

Share this project:

Updates