Vibe Coding: Real-World Example of Vibe Coding Automated Chargebacks
AI often gets a bad rap. Many don’t know how to use it properly–therefore producing incorrect responses because of user error–and several users don’t have solid use-cases. While many have caught on to AI generated content (check the em dash theory, don’t worry I’ve been using them before they were cool), the lack of creativity is what is truly holding back AI implementation. The sad reality is that those who can leverage AI the most are people that are more technical, often coders that are prone to be skeptics of AI due to the perception of AI replacing their profession. While I won’t fully argue that perspective, I don’t think there are enough real world examples of AI implementation in general business needs, so I’m going to share a simple one that has helped my company save over $2000 a month.
I have created a Python script that automatically files disputes for chargebacks at my company’s restaurant account on delivery platforms. If you don’t know how delivery platforms work, the delivery platforms will often allow customers to create a chargeback and this gets passed on to the restaurant unless contested. If a restaurant contests the chargeback, the order gets reviewed and if found to be a faulty chargeback, the platform eats the cost.
This project was completely vibe coded. I used Selenium (should have used Playwright in hindsight) Python library to facilitate clicking on my browser, and I simply select the date range of the chargebacks I want to dispute, and then let the automation script run its course. I am able to choose how many disputes I want to process and away the script goes!
With the help of AI, I am now able to save around 12 hours, along with just raw time of me forgetting to process these chargebacks. Overall I will save my company around $2000+ in both chargebacks and labor.
The point I’m trying to make here is that you don’t need to be a full-fledged coder to create new products/processes. I’ve shown you some of the backend code, and the proof of how it runs in real-time. The only thing that is holding you back is…you. Your lack of experimentation is what really holds back the usage of AI. There is also a chasm of allowing lower level users to implement AI solutions in their everyday process. I have saved my company a minimum of $24,000 a year by this implementation, but it’s only because I have a long leash. I wonder what other people can do if given this type of freedom.
Notes about the author: Traditionally an esports writer, having written over 100 articles, produced over 300 YouTube videos, multiple documentaries and several investigative pieces.
** AI Generated thumbnail, but not AI generated written words; we’re too proud for that.
*** Contact me for work if you feel you could benefit from my expertise.