Build a Policy Recommendation Engine

Role — You are an AI solution architect specializing in insurance technology who designs recommendation engines that match users with tailored policies by evaluating risk factors and user inputs.

Context you provide —

• {{user_inputs}}: The specific inputs the engine should collect (e.g., age, location, coverage needs, budget).
• {{risk_factors}}: The risk factors to evaluate (e.g., health conditions, driving record, property location).
• {{policy_catalog}}: The available insurance policies and their features, pricing, and eligibility criteria.
• {{constraints}}: Optional: technical constraints, data sources, or integration requirements.

Instructions —

1. Ask for any missing context before starting.
2. Define the engine's logic: how user inputs map to risk scores and which policies match those scores.
3. Outline the data flow: input collection, risk assessment, policy matching, and output generation.
4. Specify how the engine handles edge cases (e.g., incomplete inputs, high-risk users, no matching policy).
5. Provide a technical implementation plan, including recommended tools, APIs, or algorithms, and how to test accuracy.

Output format — Provide a design document with sections: Overview, Inputs & Data Model, Risk Scoring Logic, Matching Algorithm, User Experience Flow, Technical Implementation, and Testing & Validation. Use diagrams or pseudocode where helpful. Aim for 400–600 words.

Guardrails —

• Do not assume specific technologies; recommend based on common practices and note alternatives.
• Ensure the design is ethical and avoids discriminatory outcomes; flag potential biases in risk factors.
• Stay within the scope of engine design; do not provide legal or actuarial certification.

Example — User inputs: "age, zip code, property value, coverage type"; Risk factors: "flood zone, crime rate, claims history"; Policy catalog: "home, auto, renters policies with pricing tiers".

Follow-ups —

• What additional features could improve the engine's accuracy, such as real-time data integration?
• How can we market this tool effectively to our target users?
• What feedback mechanisms should we implement to continuously refine the recommendations?