WatchGuard Advisor
Conversational AI helping luxury watch collectors decide whether insurance makes financial sense. AI SDK + Vercel AI Gateway + Neon.
Overview
WatchGuard Advisor is a conversational AI that replaces vague insurance advice with a structured, data-driven recommendation. Through a 4–5 question conversational flow — watch value, financial picture, quoted premium, usage patterns — the advisor produces a clear verdict: insure, don't insure, or optional. Built on the Vercel AI SDK with xAI (Grok) via Vercel AI Gateway, Neon Postgres for conversation persistence, and Auth.js for user management.
Challenge
Watch insurance decisions require balancing premium cost, replacement risk, and opportunity cost in a way that most collectors don't have the financial modeling background for. The challenge was making the math feel conversational — building an advisor that feels like a knowledgeable friend rather than a spreadsheet.
Outcomes
- Conversational AI collecting structured data through natural dialogue
- Decision logic across three outcomes: insure / don't insure / optional
- Two conversation modes: MVP (efficient) and Refined (warmer, exploratory)
- AI Gateway integration enabling model switching between xAI Grok and OpenAI
- Full chat history persistence in Neon Postgres
- Guest access and authenticated flows via Auth.js