The Giftings: Building an AI-Powered Gift Matching Platform from Scratch

A case study in product design, AI integration, and full-stack development — from problem definition to a live, monetized web application.

Visit The Giftings at thegiftings.com

Role: Lead UI/UX Designer & Full-Stack Developer
Stack: React, PHP, WordPress REST API, Gemini AI, CSS Design Tokens
Timeline: Independent project, 2024–Present

1. Define: The Problem & The Goal

Gift-giving is universal — and universally frustrating. People default to generic presents, get trapped in single-retailer registries, or simply guess. There was no elegant, centralized platform that let users build a wishlist from any retailer while simultaneously capturing personal preferences to fuel intelligent recommendations.

The goal was twofold:
First, build a universal wishlist tool where users could save items from any website — not just Amazon, not just one store. Second, layer in an AI Matchmaker that analyzes who someone actually is — their job, education, hobbies, interests — and generates gift ideas that feel genuinely thoughtful rather than algorithmically bland.

A secondary goal emerged during scoping: build a robust admin console with affiliate link routing so the platform could sustain itself financially from day one.

2. Research: Understanding the Ecosystem

Before designing a single screen, I conducted competitive analysis across existing wishlist platforms. The findings were consistent: legacy platforms are functional but visually dated, locked to specific retailers, and offer zero intelligent discovery.

User interviews surfaced a core friction point I came to call discovery fatigue — the buyer’s problem, not the receiver’s. Gift givers want to give something meaningful but lack the time or insight to find it. They don’t want a link dump. They want inspiration grounded in who the person actually is.

This reframed the product entirely. The Giftings isn’t just a list tool — it’s a discovery engine for the person doing the giving.

3. Analyze: Personas & Journey Mapping

The research produced two primary personas with fundamentally different needs. The Giver is overwhelmed by choice and paralyzed by the fear of giving something wrong. The Receiver struggles to communicate their actual preferences without seeming demanding.

Journey mapping revealed a critical insight: the onboarding moment — where users input their occupation, education, and hobbies — is the entire product’s value hinge. If it feels like a form, users abandon it. If it feels like a concierge check-in at a premium hotel, they invest.

This insight drove every subsequent design decision.

4. Design: Information Architecture & Wireframing

The architecture prioritized flatness and immediacy. The homepage leads with a search bar — find a friend or family member’s wishlist without any login friction. The user profile layout gives equal visual weight to the Universal Wishlist and the AI Matchmaker, treating both as primary features rather than burying one as a secondary tab.

Wireframes started as rapid hand sketches to nail spatial relationships, then moved quickly to digital lo-fi layouts. Key decisions made at this stage: the wishlist and AI recommendations live on the same page rather than separate flows, reducing the cognitive load of switching contexts.

5. Prototype: The Design Token System & Glass-Morphism UI

Moving to high fidelity, I built a comprehensive Design Token System before writing a single production component. Spacing, typography scales, color palettes, border radii, and shadow depths are all defined as tokens — meaning the entire visual language can be updated globally from a single source of truth.

The visual identity uses glass-morphism: translucent panels with subtle light-catching borders that float above customizable background imagery. I implemented a seamless Light/Dark mode toggle with CSS animations that make the background environment subtly shift as users navigate — a small detail that makes the app feel alive rather than static.

The result is a UI that feels premium without feeling cold.

6. Test: Refining the AI Matchmaker

The most technically complex phase was tuning the Gemini AI integration. Early outputs were disappointingly generic — the model defaulted to predictable categories regardless of user input.

The breakthrough came from iterating on system prompt architecture. Instead of passing user data as a simple list, I restructured the prompt to force the model to reason about the combination of attributes — treating occupation + education + hobbies as an intersecting personality profile rather than independent data points. A “software engineer who does yoga and collects vintage jazz records” should never receive the same suggestions as a “software engineer who coaches youth soccer and restores motorcycles.”

After three rounds of prompt iteration, specificity improved dramatically. The AI began surfacing genuinely unexpected, non-obvious gift ideas — the kind of suggestion that makes a receiver feel seen.

7. Launch: The Overseer Admin Console

The final product shipped with two parallel systems: the user-facing application and a powerful Gift Overseer Admin Console built for platform management and monetization.

Through the Overseer, administrators can override global typography, inject new background imagery, adjust panel opacity, and manage the Affiliate Routing Engine — the platform’s core revenue mechanism. When users add any standard retail link to their wishlist, the system automatically intercepts and rewrites it to include designated affiliate tags. The monetization is invisible to users and seamless in experience.

The Giftings launched at thegiftings.com as a live, independently operated product — designed, built, and shipped entirely by one person.

Outcomes & Learnings

Building The Giftings end-to-end as a solo product confirmed several things: AI product quality lives or dies on prompt architecture, not model selection. Design tokens aren’t overhead — they’re the foundation that makes iteration fast. And onboarding is never a form; it’s always a first impression.

The platform is live and actively iterating. The next phase focuses on browser extension development for frictionless wishlist item capture from any retailer, and expanding the AI Matchmaker to support group gifting scenarios.