Skip to content
NR
00/Case Study

SupportIQ

AI Customer Support Platform

RoleFull-Stack Developer
Timeline2026
StackNext.js, React 19, TypeScript
SourceGitHub
Impact

Full-stack onboarding wizard with admin config and an AI-powered customer support dashboard prototype.

Overview

Context

SupportIQ is a full-stack AI Customer Support Platform built for Revolution Venture Studios. It features an end-to-end onboarding wizard with resume support, an admin panel for dynamic form configuration, and a live AI agent dashboard prototype showcasing the core product vision.

Challenge

The problem

Creating a seamless onboarding experience requires dynamic forms that can be reconfigured without code deploys. Furthermore, demonstrating the value of an AI support agent requires a realistic prototype that shows reasoning traces, real-time metrics, and ROI calculations to stakeholders.

Approach

How I built it

01

Built a 3-step onboarding wizard with session persistence, allowing users to return and resume their progress

02

Created an admin dashboard to dynamically configure which components render on each onboarding page

03

Developed a live AI customer support dashboard prototype with mock agent reasoning traces

04

Implemented real-time metric visualization including deflection rate, resolution time, and an ROI calculator

05

Designed a premium UI using Tailwind CSS v4, glassmorphism, and seamless dark/light mode switching

Technical Decisions

Why these choices

Prisma ORM with PostgreSQL

Provides type-safe database access directly from Next.js API routes, ensuring data integrity for user onboarding state and admin configurations.

Next.js 16 (React 19) App Router

Leverages the latest React paradigms for fast, server-rendered UIs while keeping backend API routes tightly coupled to the frontend in a single monorepo.

Outcomes

What shipped

Dynamic, admin-configurable onboarding flow with state persistence
Interactive dashboard prototype proving a $4.20/ticket savings ROI
Premium, accessible UI with system-preference dark mode detection
Deployed on Vercel with a Supabase PostgreSQL backend
Takeaways

What I learned

Showing an AI agent's 'reasoning trace' builds stakeholder trust much faster than a generic black-box output
Building dynamic form renderers requires strict validation to ensure at least one component renders per page