Challenge #2: PayCan - A Payment Integration Service and the Reality of AI Coding

Challenge #2: PayCan - A Payment Integration Service and the Reality of AI Coding

For my second app in this challenge, I decided to tackle a problem I've faced many times: payment integrations. The result is PayCan.

It’s like an open-source Stripe Checkout, but designed to support multiple gateways. Its goal is to help developers support multiple payment gateways through one simple integration. It's designed to be easy to connect and reliable for production environments.

The idea came from my past work. I have archited complex payment systems on two projects: the payment system for an IaaS platform and for Dasht, a no-code platform. I always wondered why there wasn't a ready-made solution that handles all your payment needs from A to Z, a service that could be connected to different apps.

This first version, v0.1, is a simple limited but working version. My main question was: how should I architect this to make it easy for developers without sacrificing security or the end-user's experience?

This project was a technical challenge for two reasons.

First, I hadn't worked with Laravel in over 10 years. The last version I used was Laravel 4, and now it's version 12. Diving back in felt like learning a new language. I had to get to know the modern PHP ecosystem, with its new features and ways of doing things.

Second, this was my first app created almost entirely from scratch with AI. I used Laravel for the backend, with Vue, Tailwind CSS, and Filament for the admin panel. My main coding partners were Claude Code and trae.ai.

At first, the AI-generated code looked perfect. It was fast and seemed to solve problems instantly. But as I went deeper into the project, many challenges appeared. The code wasn't always structured correctly for a real-world application. Sometimes it was insecure; other times, it didn't fit together well. I spent a lot of time debugging and refactoring what the AI had written. It was a powerful assistant, but it required a very experienced pilot to steer it in the right direction.

You can take an look at the project here: PayCan on GitHub (Give it a star please!)

This challenge was an intense period of learning but I liked it. I will come back to it in next sprints!

Read more

✅ The End of the Challenge: Products List and Key Learnings

Over the past few months, I set myself a bold challenge: 𝐛𝐮𝐢𝐥𝐝 𝐨𝐧𝐞 𝐰𝐨𝐫𝐤𝐢𝐧𝐠 𝐩𝐫𝐨𝐝𝐮𝐜𝐭 𝐞𝐯𝐞𝐫𝐲 𝐭𝐡𝐫𝐞𝐞 𝐰𝐞𝐞𝐤𝐬. What started as a personal experiment quickly turned into a journey of vibe coding, fast decision-making, and practicing a builder mindset. Here’s what I’ve shipped so far: 1. 𝐇𝐲𝐩– One-click installation for popular open-source apps like n8n, AnythingLLM,

By Hassan Jahan