DomainFlip AI — How I Started With an Empty Repo and Built a Full Domain Investment Platform

This is a submission for the GitHub Finish-Up-A-Thon Challenge.

I’ll keep this short since the repo docs go much deeper into the product and build. Hope you enjoy it.

What I Built

DomainFlip AI is a domain investment intelligence platform for analyzing domain quality, valuation signals, registration history, comparable sales, and acquisition potential in one workspace.

Core features

Analyze page


Enter a domain and trigger a full analysis pipeline:

• 100-point composite score across 10 weighted signals


• AI advisory layer via Gemini


• Python ML valuation reference


• RDAP registrar lookup


• Comparable sales matching with similarity scoring


• 3-year value projection with trend and risk drivers


• Deterministic Buy / Watch / Avoid investment report

Domain Comparison


Compare 2 domains side by side, or run Battle Mode with 3–5 domains to surface liquidity, brand, and acquisition leaders.

Market Intelligence page


Explore:

• TLD performance charts


• Price distribution bands


• category breakdowns


• anomaly detection for hot and soft pricing pockets


• a Domain Comparison Lab for interactive multi-domain analysis

Watchlist


A Convex-backed watchlist with target buy price, max budget, and negotiation stance tracking.

AI Assistant


A Gemini-powered assistant with Google Search grounding for domain market questions and naming suggestions.

Stack


Next.js 16 App Router · TypeScript · Tailwind CSS v4 · Convex · Clerk · Gemini (@google/genai) · Python ML · RDAP · Geist Sans/Mono

Demo

• Live App - Domain Flip AI

The Comeback Story

This project did not start with a polished product or even a clear system already in place. It started with a blank repo, a rough idea, and the challenge of turning domain investing into something that felt structured, technical, and actually useful.

Over the course of the build, it grew from a simple analyzer concept into a full domain intelligence workspace. The scoring engine came first, then valuation logic, then RDAP lookups, comparable sales, projections, AI advisory, market intelligence, watchlist tracking, and comparison tools. Each layer made the product feel less like a demo and more like a real decision-making surface.

What I’m proud of is that the project didn’t stay at the level of “enter a domain, get a number.” It became a system that tries to explain why a domain looks strong or weak, what the risks are, what the market context looks like, and how someone might actually act on that information.

By the end, it felt less like I had built a single feature and more like I had built the foundation of a real product.

My Experience with GitHub Copilot

GitHub Copilot helped most with the repetitive parts of the build: boilerplate types, repetitive JSX structures, utility scaffolding, and other verbose implementation work where the pattern was already clear.

It was especially useful in places where the logic was straightforward but time-consuming to type out, like cache helpers, comparison table structures, and fallback object shapes for the AI advisory layer.

Overall, it felt less like a replacement for decision-making and more like a speed multiplier. I still drove the architecture and product direction, but Copilot helped reduce time spent on mechanical implementation.