AI-Powered Platform for Short-Term Personal Property Insurance

The platform served a dual purpose:

For customers: A user-friendly mobile and desktop app for selecting and purchasing insurance policies, tailored by coverage period, risk scope, item type, and insurance history.

For underwriters: A data-backed toolkit to test new policy models with minimal exposure, allowing them to validate hypotheses, optimize pricing, and forecast insurance reserves.

The client initially contacted Devox Software to invite us into the competition. After considering the cost estimate and value proposition, they chose us because we offered 30% faster development.

We were asked to launch the pilot product to validate market interest and estimate potential insurance reserves for future scaling. That’s why the MVP focused on capturing real-time KPIs, loss tracking, and testing underwriting hypotheses in a live environment.

A distributed team of 20 professionals across New York and Europe collaborated remotely. The team covered:

• Backend and Database Development
• Frontend and Mobile App Development
• Data Science & Machine Learning
• Product and Project Management
• UX/UI and KPI Dashboard Design

under the solid management of a PM with a track record in insurtech.

• No standard libraries for insurance modeling or rating logic were present,
• A complex tariff engine was needed to support highly customizable policies (e.g., insurance starting from a single day),
• Dynamic dashboards to track losses, underwriting metrics, and business KPIs in real-time that served as a basis for strategic decisions,
• Strict compliance regulations in finance, insurance, and personal data,
• Implementing insurance history and scoring techniques was necessary to prevent fraud and mitigate losses,
• The project faced limited timeframes, requiring the development of both a web app and a desktop version within a short period.,
• The need to span across the insurance full cycle, from scoring and underwriting to claim management and indemnification.

Languages: Python, SQL

Frameworks & Libraries: TensorFlow, Scikit-learn, Pandas, NumPy

Backend: Django, Flask

Frontend: HTML5, CSS3, JavaScript

Dashboards & Visualization: Dash, Plotly, Grafana

Database: PostgreSQL

Infrastructure: AWS (EC2, S3), Docker

Monitoring & Logging: Prometheus, ELK Stack

We’ve delivered an MVP just within 3 months from scratch, covering:

• Sleek Front End: Appealing interfaces with robust UX/UI, user onboarding, and technical architecture.
• Solid Security: Compliance under financial and data protection regulations, including seamless policy creation and payment with auto-renewal options,
• Fraud Prevention and Risk Mitigation: Auto-scoring procedures and loss history evaluation to prevent insurance scams and fraud,
• Dynamic Pricing: Flexible tariff engine with risk-based pricing,
• Real-Time Reporting and Forecasting: Underwriter dashboards for monitoring KPIs, loss ratios, and claims to enable instant business decisions, based on AI modules for user behavior analysis and pricing adjustment,
• Full-Cycle Insurance: In-built claim and indemnification modules represented a structured one-funnel approach in management.

The system scaled to serve 5,000–10,000 active users in the pilot stage. Then, the platform has quickly acquired 50k+ users across numerous states. The client has reported the following achievements:

• Fully Functional MVP Delivered: The platform was built from scratch and successfully launched, providing both user-facing features and underwriter tools in a single, cohesive product.
• Validated Market Hypotheses Through Real-Time User Data:  The pilot has generated valuable behavioral and usage data, allowing to test pricing assumptions, understand customer preferences, and predict losses.
• Scalable Architecture Enabled by AI-Driven Insights: With machine learning modules integrated from the start, the system laid the groundwork for scalable, intelligent decision-making in real time.
• Deep Understanding of Loss Dynamics and Risk Modeling: The platform’s custom dashboards provided underwriters with actionable insights into loss ratios, claim behavior, and policy performance. This information gave them the room to adjust reserve strategies and confidently plan for future product iterations and market expansion.