I Built 10 ZK Demos in 2 Weeks: A New Era of "Vibe Coding" with AI + ZKPs



I’ve been troubled lately. The zero-knowledge industry is flooded with brilliant academic research, yet I see a frustrating scarcity of practical, user-facing applications for Zero-Knowledge Proofs (ZKPs). What are we actually building with this incredible technology?
Depressed by this gap but also inspired, I decided to dive in and build them myself. Thanks to my very capable product management team at Horizen Labs, I was able to embark on a personal challenge. In just the last two weeks, I’ve built 10 working demos of ZKPs in practical, end-user applications—a journey I believe few VPs of Product have taken on personally.
This series is the story of that journey. It’s a story about "vibe coding" and innovation, where I paired my ideas with powerful AI tools like Gemini to bring complex applications to life. My process was simple: I knew what I wanted, I specified it, and I let the AI handle the heavy lifting. Some might say that’s what makes a good product manager!
I started with a classic academic example—proving knowledge of a number's factors without revealing them—using Circom for the circuit and a simple UI. I quickly realized I had done very little actual work on my own and this was the tremendous value of vibe coding.
That’s when I got bitten by the "vibe coding" bug and decided to build something with real utility. Being a fan of horse racing and sports, I naturally gravitated toward gambling.
In the upcoming posts, I’ll take you through my experiments:
- Provably Fair Blackjack: I’ll detail my first foray into building a game of chance where ZK proofs guarantee the integrity of the shuffled deck. We’ll explore the back-and-forth with Gemini to debug the code and the critical user experience challenges I faced, like a 30-second verification time using zkVerify that’s just not acceptable for a fast-paced game.
- On-Chain Slot Machines: Follow along as I build a provably fair slot machine. Frustrated by the classic casino experience of quick wins followed by bigger losses, I wanted to design a game where players could be certain of its integrity. This post will cover how I used Gemini to create the Circom circuit and integrated a simple on-chain economy on Base, allowing players to buy and burn ERC-20 tokens to play.
- Many other use cases: The real breakthrough came when I moved beyond Circom and started using zkVMs and an innovative product that we’re incubating at Horizen Labs. This enabled even faster product development, with near-instantaneous verification times and endless possible zk-enabled products.
This series is for anyone curious about the practical application of ZKPs and the power of AI-assisted development. Follow along as I share the code, the challenges, and the solutions I found while bringing these demos to life.
If you're into ZK, AI, and building real things—not just reading papers—follow me on X: @john_camardo.