Recourse.ai - High-Fidelity Parametric Insurance Prototype

Recourse.ai is an Android prototype that bridges AI operational telemetry with parametric insurance workflows. Designed for the Lloyd’s Lab ecosystem, it demonstrates how organizations can bound the downside of automation by converting real-time agent behavior into instant, verifiable insurance payouts.

---

📖 Table of Contents

1. Business Case
2. Target Personas & Use Cases
3. Application Preview
4. Architecture
5. Technology Stack
6. Project Structure
7. Getting Started
8. Configuration
9. Security & Compliance
10. Roadmap & Venture Strategy
11. Contribution & License

---

💼 Business Case

The Problem

Traditional Errors & Omissions (E&O) insurance is ill-equipped for the "speed of AI." Claims take 90+ days to settle, underwriting is based on static forms rather than live telemetry, and "Silent AI" exposure creates unquantifiable risk for the board.

The Recourse Solution

Recourse.ai provides a Continuous Monitoring Layer that:

• Quantifies Risk: Assigns a "Guardrail Maturity Score" based on live agent behavior.
• Automates Payouts: Uses parametric triggers to clear funds in < 48 hours when pre-agreed safety thresholds are crossed.
• Benchmarks Performance: Provides CFOs with real-time ROI and efficiency gain metrics relative to traditional benchmarks.

---

👥 Target Personas & Use Cases

Persona	Primary Objective	Key Feature Used
CFO / COO	ROI & Financial Stability	Executive Metric Grid (Recovered Funds, Efficiency Gains)
Head of AI	Operational Reliability	Agent Telemetry & Automatic Safety Kill-Switches
General Counsel	Governance & Audit	Region-specific Compliance Audits (EU AI Act, DIFC)
Broker	Capacity Distribution	Capacity Monitoring & Parametric Policy Management

---

📱 Application Preview

1. Unified Executive Entry

Secure, persona-driven authentication designed for instant boardroom demonstrations.

2. Operational Risk Dashboard

A real-time overview of AI agent health, guardrail maturity, and anomalous activity pulses.

3. Incident Deep-Dive

Automated risk assessment of flagged anomalies, including estimated financial exposure and jurisdictional legal context.

4. Parametric Policy Logic

Full transparency into the "Rules Engine" logic that governs automatic insurance triggers.

---

🏗 Architecture

High-Level System Design

The prototype uses a decoupled MVVM architecture with a resilient data layer that supports both live JDBC connections and high-fidelity mock fallbacks for offline stability.

graph TD
    User((Executive Persona)) --> UI[Material 3 Frontend]
    
    subgraph "Android Application"
        UI --> VM[MVVM ViewModels]
        VM --> State[StateFlow / UI State]
        VM --> Repo[Repositories]
        Repo --> Logic[Business Logic / Payout Engine]
    end
    
    subgraph "Data Persistence"
        Repo --> DBH[DatabaseHelper]
        DBH -- "JDBC (Async)" --> MySQL[(MySQL 8.0)]
        DBH -- "Local Mode" --> Mock[Strategic Mock Data]
    end
    
    MySQL -.-> Schema[9-Table Insurance Schema]

Loading

Data Flow Lifecycle

The following diagram illustrates the request-response lifecycle during a parametric trigger verification:

sequenceDiagram
    participant U as User
    participant V as View (Activity)
    participant VM as ViewModel
    participant R as Repository
    participant DB as MySQL DB
    
    U->>V: Select Incident
    V->>VM: Request Detail(incidentId)
    VM->>R: Fetch Telemetry & Thresholds
    R->>DB: SQL Query (JOIN agents/policies)
    alt Database Connected
        DB-->>R: ResultSet
    else Local Mode (Fallback)
        R-->>R: Load Mock Strategic Data
    end
    R-->>VM: Incident Domain Object
    VM-->>V: Update UI State (Success)
    V-->>U: Display Risk Report & Claim Status

Loading

---

🛠 Technology Stack

📱 Frontend / Mobile

• Kotlin: Primary programming language for modern, type-safe development.
• Jetpack KTX: Used for concise, idiomatic Android development (e.g., isVisible).
• Material 3: Premium UI components and Navy/Teal branding.
• Coroutines & StateFlow: Asynchronous stream handling for high-performance UI updates.

🗄 Backend & Data

• MySQL 8.0: Relational database for policy, claim, and telemetry storage.
• JDBC: Direct database integration for low-latency prototype response.
• Resiliency Layer: Custom "Local Mode" logic that ensures 100% uptime during pitches.

🛡 Security & Governance

• Compliance Engines: Hardcoded logic for EU AI Act (Art. 14), DIFC Reg 10, and MAS Sandbox Plus.
• Parametric Verification: Deterministic rules-engine logic to prevent "Black Box" claim disputes.

---

📁 Project Structure

Recourse/
├── app/                        # Main Android module
│   ├── src/main/java/          # Kotlin source code
│   │   ├── data/               # Data Layer (Models, Repositories, DB)
│   │   └── ui/                 # UI Layer (Activities, Adapters, ViewModels)
│   └── src/main/res/           # Android resources (Layouts, Themes, Values)
├── docs/                       # Project Documentation
│   ├── screenshots/            # Visual assets for README
├── README.md                   # Enterprise documentation
└── build.gradle.kts            # Project-level build configuration

---

🚀 Getting Started

Prerequisites

• Android Studio Ladybug or newer.
• JDK 11 (Standardized target).
• MySQL Server 8.0 (Optional, app will run in "Local Mode" without it).
• Pixel 8 Emulator/Device (Recommended for layout integrity).

Installation

1. Clone the repository:

   git clone https://github.com/alfinohatta/recourse.git

2. Open the project in Android Studio.
3. Copy .env.example to .env (Symbolic for prototype config).
4. Sync Gradle and build the project.

Running the Demo

For the "One-Tap Strategic Demo":

1. Deploy the app to an emulator.
2. On the Login screen, enter: dana.whitfield@northwindapparel.example.com.
3. Tap SIGN IN. The app will bypass DB latency and load the full CFO dashboard immediately.

---

🛡 Security & Compliance

• Data Minimization: The telemetry layer only captures agent metadata and financial amounts, avoiding PII where possible.
• Local Resilience: In a production environment, JDBC would be replaced by an authenticated REST API. This prototype uses direct JDBC for performance during high-stakes demonstrations.
• Compliance Audit: View the "Risk Report" in any AI Agent detail view to see the automated audit trail for regional regulations.

---

📈 Roadmap & Venture Strategy

Recourse.ai is positioned as a first-mover in the Agentic AI Insurance space. Our multi-year sequencing includes:

1. Phase 1: Beachhead in D2C Retail ( MAS/ADGM Sandboxes).
2. Phase 2: secure A-Rated Capacity & Formalize Broker Channel.
3. Phase 4: Pivot to embedded platform distribution (White-label).

---

📄 Contribution & License

This project is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0).

Why AGPL?

We believe in the transparency of insurance logic. If you use the Recourse Parametric Rules Engine in a hosted service, you must share your modifications with the community to ensure "Black Box" insurance remains a thing of the past.

---

Trusted by Lloyd's Lab Alumni | Bounding the Downside of Automation