For those not keeping up to date with their favourite dictionary’s word of the year, Collin’s has recently announced that “vibe coding” is their contribution for 2025 (forgetting the fact it’s two words, along with another nominee of “aura farming”). Defined as “the use of artificial intelligence prompted by natural language to assist with the writing of computer code”, vibe coding has hit the software industry as a contentious topic, leading to more heated discussions than the age-old “tabs vs spaces” arguments.
What the definition misses is the cultural usage of the practice - typically, vibe coding involves providing prompts to your preferred AI (the jury is still out), letting it develop the feature, ensuring the output works at a surface level, and then moving onto the next task. Sounds great on paper, it works! But in reality, this practice can be far more detrimental for businesses in the long run.
Paytently prides itself on the resilience and scalability of our APIs - we’re a payment gateway, if we can’t process payments quickly and all the time then we’re not doing our job right. However, to build resilient and scalable systems we need our engineers to have a deep understanding of the code that they’re writing. Vibe coding lends itself towards having a lax attitude in the code that is being developed; sure, the feature might work, but I can almost guarantee you that there’s an edge case that’s been missed.
Fundamentally a Large Language Model will not have the domain knowledge at both a technical and product level of the business’s systems - having skilled engineers who understand the domain not only have the context of what feature is being created at that point in time, but also how that system will be growing in the future.
Additionally, our engineering team believes that the best way to learn is by creating; actually writing code, understanding it line by line and knowing in-depth how the entire system works. Sure, you can vibe code a feature into an existing codebase and read what was created, but like anything else in life, how much of that will sink in? You can listen to a melody played on an instrument a hundred times over, but playing that melody yourself gives you a much greater understanding.
Now all this isn’t to say that Paytently is anti-AI, far from it! We utilise coding agents on a daily basis to help provide boiler plate code, write tests faster and even assist with our reviews. AI is also a key component of our payment routing systems, turning it into a dynamic, intelligent, and highly adaptive process. This technology enables payments to be routed in real-time without human intervention and with remarkable efficiency and accuracy, selecting the optimal path in just milliseconds and increasing acceptance rates across the board!
If you’re a hobbyist looking to make a quick prototype, or you don’t know how to code and want to make a (maybe somewhat unreliable) project, vibe coding is a great practice to leverage. But if you’re building highly performant, highly reliable and highly scalable systems, like we do here at Paytently, it’s best to keep the vibes for your after-work drinks
Written by: Chawin Tree
.png)
.png)
.png)
