Building a Web App MVP in 24 Hours with AI & Next.js

As a Web Developer at Ipanova, I love tackling technical challenges. It all started with a call from an old client who had built a React app for wine tasting as part of a wine tasting experience box. Their app was still running, but they were looking into a new B2B business model powered by AI.

The Unexpected Callback

When I first reached out to catch up, the answer was expected: no budget for their current app's evolution. Instead, they had a new idea: a PWA (Progressive Web App) aimed at delivering an ultra-immersive experience, leveraging AI without going through the App Stores.

Their technical co-founder was already using Next.js and AI tools like Cursor — my preferred stack. The call ended with the usual "We'll handle the MVP internally for now." No surprise. Still, I was glad to reconnect.

A Weekend Later… Change of Plans!

Monday morning, I get a call back. "Hey, let's set up a scoping workshop and get a quote for the MVP!" The door was open.

The Challenge: Build a Working MVP Demo in 24 Hours

From past experience as both a developer and former sales rep, I knew something crucial: if the client can see and interact with a real prototype, they will be much more likely to commit to the project.

Why not challenge myself to build a working MVP demo in just one day? Not wireframes — a real clickable, testable version. Two purposes:

1. Give the client a real feel of the application.
2. Ensure we gather solid requirements in the process.

The Tech Stack: AI-Powered Development

My secret weapons: Cursor, Next.js, ShadcnUI, and Vercel AI SDK.

1. Defining Clear Rules & Context

Cursor allows developers to define rules in .mdc files so the AI understands the syntax, frameworks, and conventions in use.

2. Structured AI Prompts for Each Feature

Instead of just "asking AI to code", I carefully prepared prompts for every feature providing feature details, user interactions, and UI behaviour. Each prompt acted as a blueprint for efficient AI-assisted pair programming.

3. Rapid AI-Powered Development

Armed with ChatGPT and Cursor, I iterated fast. The AI handled repetitive coding while I fine-tuned logic, integrations, and UI tweaks. The highlight: integrating an OpenAI-powered chat feature with Next.js and Vercel AI SDK — surprisingly simple, with outstanding results.

The Final Touch

After a day of coding I fixed bugs, polished interactions, and prepared a structured demo walkthrough. The biggest takeaway wasn't just how fast AI can boost development, but how critical well-defined specs and user projection are for any project.

Lessons Learned

1. AI-assisted development is a game changer if you know how to guide it.
2. A structured approach (rules, prompts, context) leads to significantly better output.
3. Clients love seeing tangible results — not just pitch decks.
4. Having a technical conversation with someone using the same stack is rare and priceless.

Taking risks and pushing for innovative approaches is what makes this job so exciting. 🚀