Of course I jumped into the Vibe Coding Trend
Right after I jumped into the MCP Servers Trend
As I shared a few months ago, I joined Kernel, an amazing fellowship for peer-to-peer learning focused on Web 3, sustainable development, and a more conscious approach to technology. During this journey and in a very organic way I developed a project. This was not my initial purpose as I joined the program. However, during the process, I realized this was the best way to take the most out of the program and my peers' knowledge. The environment around Kernel is so welcoming that I just wanted to make myself useful, so what I thought as a project was in a space I feel very comfortable sharing: Product and Growth Strategy.
The initial idea was to build a tool that provides a companion (Copilot) to product builders, and founders to keep track of their engagement and revenue metrics.
The challenge was jumping from the theory to build an actual tool. As many of you know, I am a black belt in jumping into trends. So I began doing research about MCP Servers, and the best Vibecode tools for building “A Growth Copilot”.
This and the following Substack I’ll be talking about my journey using Vibecoding tools and MCP Servers for building what ended up being, an education game for Product and Growth Strategy.
For making a simple validation of what I was trying to build I connected a simple MCP Server using the Python Public SDK for a game called Two Truths and a Twist, shout out to Ergodic Labs for making this repo available to curious people. The logic of the server is simple it turns your Claude Copilot into a trivia game, I use this during my office hours to make some trivia rounds about product strategy in Web 3, and ended up being super engaging. So for the MVP, I defined the:
An external UI for Web and Mobile.
User Auth and log in to keep track of user engagement and token usage.
And a leaderboard.
The Stack I Used for this was:
What it is: A chat-based AI assistant for web development that generates code using modern technologies like React, Next.js, and Tailwind CSS. I used v0.dev to quickly prototype the game's UI and integrate client-side functionality. Between setting up the UI and connecting the endpoints I spent around 3 hours. It is a Vercel tool so it is very simple to deploy into production.
What it is: An open-source backend-as-a-service platform for databases, authentication, and APIs. Supabase served as the backend for my game, handling user authentication and database management, it has native integration with v0.dev, so I was able to set up every required table using natural language.
What it is: An AI-powered code editor based on Visual Studio Code, offering intelligent code suggestions and dynamic optimization. Cursor helped streamline my coding process with advanced autocompletion, error detection, and multi-line edits. Cursor is a very powerful copilot to write code.
For the Beta Version I want:
Enable Users to save their game streak
Select Categories for the Trivia Game
At some point I want businesses or nodes to be able to upload their database and run their trivia game based on this info, for this, I need to create a backend in Python. Still, no need to update the current tool stack.
There are two Web 3 Vibe Coding tools I want to explore and I am still figuring out if there is a use case for this project or if I need to think of it for the next one, but Web 3 Enterprise Solutions company Alchemy launched its copilot for completing a smart contract in ETH and TON ChatWeb3, and Jit Codes is a Solidity Code Generator.
I think there is a trend for builders to adopt this tool stack, in the validation and user adoption process the game is about fast iteration.
The adoption of vibe coding tools has skyrocketed in recent years:
By 2023, 44% of developers were already using AI coding tools like GitHub Copilot
GitHub Copilot surpassed one million users by 2025, while Replit reported that 75% of its AI-enabled users relied heavily on AI for coding tasks.
Startups have embraced vibe coding for rapid MVP development. For instance, nearly 25% of Y Combinator startups now rely on AI-generated codebases.
The Tradeoffs of Vibe Coding
As with any other technology adoption, it brings new tradeoffs. I’ve been hearing some discussions about security risks. New attack vectors like prompt injections are emerging alongside these tools. Also just copying a public repository and connecting it to your local projects opens tons of risky gates. On the social side, the rise of vibe coding highlights an abundance of "coding work labor," where repetitive tasks are automated by AI. However, this trend raises critical concerns:
AI-generated code often lacks robust security measures such as input validation or encryption. This can introduce vulnerabilities that developers may overlook during rapid prototyping.
Scaling AI-generated projects often requires significant human intervention to refine backend logic or ensure compliance with industry standards.
While vibe coding democratizes software creation, it risks sidelining engineers focused on foundational aspects like infrastructure and security. The industry must balance innovation with cultivating expertise in these critical areas.
In this era of rapid innovation, we must remain vigilant about the tradeoffs between speed and quality. Vibe coding is undoubtedly an enabler for people like me, I think is useful for simple validations, MVPs, and stand-alone features. But should complement not replace the engineering practices that underpin secure and scalable systems.
You know this Substack aims to be a playful space as both an insightful exploration of your experience and a thought-provoking commentary on the future of software development, crypto degens, and worried finance bros!