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

What I Built

PharmVerify NG is an AI-powered drug authenticity and alternatives checker built specifically for Nigerian patients. It lets you search any Nigerian drug name and get back real, structured information — NAFDAC registration numbers, manufacturer details, verified generic alternatives, and direct NAFDAC verification links — all powered by the Claude AI API.
🔗 Live app: https://pharm-assist-tool.vercel.app
💻 GitHub: Torisbrain/pharm-assist-tool
This project started as part of my participation in the Qubators AI Foundry Cohort 01 — a 30-day fellowship where fellows are expected to ship a real, working AI product from scratch, with equity terms and a signed MOU. No pressure, right?
The idea came from a real frustration I've watched people around me experience: buying medication in Nigeria without any easy way to verify if it's genuine, check the NAFDAC number, or know if a cheaper generic alternative exists. Fake and substandard drugs are a serious problem in Nigeria — and most patients have no accessible tool to protect themselves.
I wanted to build something that actually helps.

Demo

<Live URL: https://pharm-assist-tool.vercel.app
The app has 4 live pages:

Home — Search bar with instant drug lookup
Results — AI-generated drug profile with NAFDAC number, manufacturer, and verification link
Alternatives — Generic and equivalent drug suggestions
About — Mission and usage guide

Here's what a typical flow looks like:

User types a drug name (e.g., "Lonart DS" or "Amoxil 500mg")
Claude AI processes the query against Nigerian pharmaceutical knowledge
User gets back: NAFDAC number, manufacturer, authenticity notes, and generic alternatives
One-click link to verify directly on the NAFDAC portal

The app currently returns results for every Nigerian drug queried, with a 0% error rate on build and deployment.

The Comeback Story

Where It Started
The concept began as AuraBridge Health / Aivara Care — a broad, ambitious AI pharmacy and clinical data platform I was conceptualizing. It was a big idea with a lot of moving parts and no clear MVP scope. I had notes, rough wireframes, and a vague notion of "AI + Nigerian healthcare" — but nothing shipped.
Then Qubators happened.
When I joined the AI Foundry Cohort, the clock started ticking. 30 days. Real product. Real users. The broad concept had to be stripped down to something I could actually finish. I narrowed it from a full clinical platform to a single, high-value action: let any Nigerian patient verify a drug in under 30 seconds.
What Changed
BeforeAfterVague "AI health platform" conceptFocused drug authenticity checkerNo code, no repoDeployed to Vercel, auto-deploys on pushNo AI integrationClaude API fully connected and returning real results0 pages built4 pages live and functionalUndefined tech stackTanStack Start + Vite + Tailwind + Claude APINo user flowEnd-to-end: search → result → verify → alternatives
The biggest shift wasn't technical — it was mental. I stopped trying to build everything and started asking: "What is the one thing that makes this worth existing today?"
The answer was clear: drug verification for Nigerian patients, powered by AI, accessible from any phone.
Once I locked that in, execution became much faster. I used Lovable to scaffold the frontend quickly, connected the Claude API as the intelligence layer, set up Vercel for deployment, and pushed the repo to GitHub for version control and CI/CD via auto-deploy.
Every drug search now works. The build passes clean. The product is live.

My Experience with GitHub Copilot

GitHub Copilot was genuinely useful throughout this project — especially during the parts where I was moving fast and couldn't afford to slow down for boilerplate.
Where Copilot helped most:

1. API integration scaffolding When I was wiring up the Claude API calls, Copilot auto-completed the fetch request structure, headers, and response parsing almost perfectly. What would have taken me 20–30 minutes of looking up documentation took about 5. It understood the pattern from the first few lines I typed.
2. Tailwind class suggestions Building the UI quickly with Tailwind meant typing a lot of utility classes. Copilot's inline suggestions for responsive layout classes (sm:, md:, flex, gap-) saved real time and kept the design consistent without me context-switching to docs constantly.
3. TypeScript type inference The project uses TypeScript, and Copilot helped me correctly type the API response objects from Claude without having to write the interfaces entirely from scratch. It inferred shapes from how I was using the data and suggested accurate type definitions.
4. Debugging assistance When I ran into an issue with how the API response was being parsed (content blocks from Claude can be mixed types), Copilot suggested the correct .filter() + .map() pattern to extract only text blocks — which fixed it immediately. What I'd tell other builders: Copilot doesn't write your product. But it removes the friction between knowing what you want to build and actually building it. For a 30-day sprint with a real deadline, that friction removal is everything.Tech Stack

Frontend: TanStack Start, Vite, Tailwind CSS
AI Layer: Claude API (Anthropic)
Deployment: Vercel (auto-deploy on GitHub push)
Version Control: GitHub (Torisbrain/pharm-assist-tool)
Dev Tools: Cursor, VS Code, GitHub Copilot

<! What's Next
PharmVerify NG is live but the roadmap is real:

NAFDAC database integration for direct cross-referencing
Drug interaction checker
Pharmacy locator (find where to buy verified drugs nearby)
Offline-capable PWA for low-connectivity areas
Hausa and Yoruba language support

Final Thoughts
The hardest part of finishing a project isn't the code. It's choosing what the project actually is — and being willing to ship a smaller, better version instead of waiting for the perfect one.
PharmVerify NG went from "a vague idea about AI and Nigerian healthcare" to a live, working product that any Nigerian patient can use today. It's not everything I originally imagined. It's better — because it actually exists.

Try PharmVerify NG live →
GitHub repo →https://github.com/Torisbrain/v0-aura-pharm-app