Keep up to date with the trending topics
from our industry experts

Everyone can be a coder with AI, but should they?

Written by Dan O’Hara

Tuesday, 19 August, 2025

Webinar:

The true cost of AI: A reality check for fund managers

On-demand

AI now comes with a meter running. Join our expert panel to explore how consumption-based pricing is reshaping AI adoption and how to prove it’s delivering measurable value.

With the advent of vibe coding, developing apps within financial services comes with a big risk

If there is one thing AI is fairly good at, it's computer programming. With the vast amount of online resources available for training and combined with the fact that the underlying source code to apps, websites, scripts etc, are just all text, it’s no surprise that one of AI’s most prevalent use cases is supporting the software development lifecycle. We are already seeing major software development houses like Microsoft and Salesforce reduce parts of their workforce, as AI drives significant efficiencies in their software development process, by churning out code quicker than a human ever could. But amid all this excitement, is there anything financial services firms should be concerned about?

What’s vibe coding?

A new term has recently entered the tech vernacular: vibe coding. It describes a new type of development process, where the AI takes on the role of the developer, and you act as its customer, simply explaining what you want to achieve, without needing to write any computer code yourself.

This approach is empowering anyone to start building websites and mobile apps from scratch, just by describing how they should behave, what features they should include, and how they should look. More and more stories are emerging of startups launching with a fully-fledged website or app that has been built by people who have never written a line of computer code before.

Does vibe coding actually work?

Yes, if you ask an AI to write code, in virtually any computer language, whether it’s Python, JavaScript, HTML, or Objective C, it will quite happily reply with the generated code. In fact, there are even some AI tools now designed specifically so you don’t even need to see the code at all. Instead, they will show you a preview of your app directly within the interface, dynamically updating it as you provide new updates.

But, this convenience comes with a considerable warning: just like with all AI, you can’t always trust the results. You only get out of it, largely, what you put in, via the quality of your prompt, and the data the AI is using as its knowledge. An AI will always respond with something, but that doesn’t necessarily guarantee it’s the right way to do something or the way a professional developer would approach the same task.

For example, it might be easy to ask, “Write an iPhone app that connects to FactSet and displays current and historical prices for our securities.” But the real question is: “Is it actually safe and secure?”

What are the risks for financial services?

Whilst it’s very tempting to go straight from the drawing board of an idea to “vibe coding” a real-life app, the reality is that professional developers, ideally more than one, should still be involved in the software development process. In highly regulated industries like financial services, where security is paramount, AI is just not at the level where it can be trusted to deliver a secure, hardened software project on its own. Because AI-generated code depends on how a prompt is written, it cannot reliably follow security best practices, anticipate the security implications of design decisions, or validate the security of its choice of code, at least not in the way a human could.

Whilst humans can make mistakes too, years of evolution in the software development lifecycle have created robust processes to mitigate those errors, and those safeguard processes just aren’t in place with vibe coding.

Is there a place for vibe coding?

AI still plays a valuable role in the software development process. In fact, it’s become quite common for developers to use AI to make an initial start on code or fine-tune their existing code for improvements. For example, Microsoft regularly uses AI to perform initial code reviews, where AI checks written code published by developers for obvious mistakes that would prevent it from functioning correctly.

So yes, there absolutely is a place for AI in app development, but for now, it definitely should remain in the co-pilot seat, assisting rather than fully taking control!

If you are interested in learning more, then please speak to one of our experts.

Our services

Our 4 pillars for a complete choice of managed IT services - all tailored to the needs of financial services firms in London and the UK.

Finance Forward 365

Microsoft modern workplace & cloud technology for digital transformation.

Compliant Teams

Increased productivity & collaboration with call recording whilst reducing costs.

Cyber Security

Keep your data secured against rapidly changing threats within Financial Services.

Power BI

Business Intelligence transformation and support tailored for Financial Services.

The LanWIRE

Join the community for financial services businesses

  • Stay updated with technology and cyber security trends
  • Network with your industry peers
  • Get invites to webinars and exclusive events
  • Gain access to useful tools and templates

The LanWIRE

Join the community for financial services businesses moving to the Microsoft Cloud

Q