Posted by Eva Zarai. Last updated: May 14, 2026
For the less tech-savvy, the indoor smoking ban was probably the most important vibe-coding event in the currency market over the past 20 years. Until now. 
Dutch bank ING has expanded its use of AI and started experimenting with so-called vibecoding. This is a technique that uses natural language to prompt machines to write code and create complex systems, following prompts from humans, rather than using specialized knowledge. A bit like Star Trek, but for trading.
This rapidly spreading practice is more than just hype, it has the potential to disrupt the relationship between established technology providers and banks, and could soon allow third-party companies to perform many of the services they provide for much less. “I think this is a big risk for providers. Banks are always trying to cut costs, and the savings add up to be significant,” said Simon Bevan, global head of electronic trading at ING. complete fx.
ING has so far found two main use cases for vibe coding. One is on the trading side, where quantitative desks can write code and create test environments for it within hours rather than days or weeks. There’s no need to spend time on experts, senior developers, or expensive developers. The second is visualization.
“At this point, our quant desks are using Vibecoding on an almost daily basis,” Bevan said, adding that deployment so far has only been done under strict supervision, with senior team members overseeing the output. However, progress is rapid and ‘vibing’ already promises significant cost savings for users.
Bevan points out that FX needs the best developers. Because banks’ success or failure depends on their developers’ ability to generate extremely reliable code with incredibly low latency that can process huge amounts of data. FX makes big decisions and takes on big risks for banks, so it needs to be very reliable. This is not cheap. As more junior staff members compete at this level, senior developers are freed up to focus on more creative efforts.
“I was surprised at how good the results turned out,” he says. “One of the first use cases we found was writing code and writing a testing framework for it. Instead of taking days or weeks, all of a sudden you could do it in hours. All of a sudden, all of our amazing developers were freed up to do more valuable work.”
“This is similar to how AI is used to free up time to do what humans are good at: being creative,” he added.
Another practical use case is visualization. Turning market data into visually appealing products required UX specialists and third-party systems, but with vibing we can easily and quickly do this in-house. And while hallucinations and other flaws can be a risk for traders, Bevan says the risk and reward of visualization is very attractive.
“We are in an experiment to see if we can replace expensive visualization tools with our own in-house builds,” he explains. “This is a potential cost savings, but it goes beyond that. Because we do it in-house, we can make it very bespoke, so not only do we save money, but we can improve what we do, we don’t have to hire very expensive UX people, and we can do it with more junior talent. It’s not perfect yet, but it doesn’t require any major adjustments.”
Using an in-house human model reduces privacy concerns as no data leaves ING. And applications are emerging rapidly. ING Financial Markets began experimenting with Vibecoding in just February, but after a short trial, the potential quickly materialized and the use cases snowballed.
Progress has been so rapid that Bevan warns that rules and regulations need to catch up. While there are frameworks for using AI for clients, they do not yet exist for trading or other areas. And oversight remains important, especially on the transactional side.
However, given the potential to reduce costs and improve banking services, the exuberance is unlikely to fade away. Bevan said that for sales and trading-side applications, teams may use tools that allow them to simply type in a type prompt to generate bespoke reports about customers, their trading patterns and hit rates that suit their presentation.
“The speed is insane. We’re seeing progress every week. I think within a year it’s going to be ubiquitous and I think every bank will have it, but it’s going to work at a different speed. Banks that can’t get on board with this will be left behind,” Bevan said.
