What is vibe coding?

The term vibe coding It emerged in early 2025 and has been an integral part of our lives ever since. The idea is simple: you describe what you want in plain language, and AI writes the code. You evaluate the result, not the code itself. If it doesn’t work as you’d like, you simply adjust it with a new prompt.

What used to be a project for developers can now be done by a staff member in an afternoon. Think of a dashboard for their own department, a chatbot that answers colleagues’ questions, or a small app that replaces an Excel process. Tools such as Microsoft Copilot, Copilot Studio and Power Apps make this very accessible within Microsoft 365. What’s more, new AI tools promising the same thing are popping up every month.

Why this isn’t just hype

The time it takes to turn an idea into a working tool is shrinking from weeks to hours. And the people who know the process best can now build it themselves. This takes the pressure off IT, speeds up business operations and delivers solutions that truly align with the way people work.

My belief is that organisations that embrace this will gain momentum. Organisations that ban it will find that it happens anyway – only behind the scenes.

The risk: shadow IT 2.0

Because that’s the flip side. Every employee who builds is also an employee who makes decisions about data, access and security. Often without realising it. We’ve already seen this with shadow AI: if there is no safe alternative, people will find their own way.

• Data in the wrong place: customer or staff data that ends up in a public AI tool.
• Access that is too unrestricted: an app that accesses or shares more data than necessary.
• Nobody knows it exists: tools that are gradually becoming business-critical, without IT realising it.
• No longer the owner: the developer leaves or changes roles, and nobody is maintaining the solution any more.

Bans don’t work, but guidelines do

The solution is not to put the brakes on, but to create a safe environment. An environment in which staff can experiment freely, within clear boundaries. In practical terms, this boils down to six choices:

• Choose a platform and open it. Provide staff with an official space to build, preferably within your existing Microsoft 365 environment, using the same identities and security measures.
• Set out the rules within the platform itself. Use separate environments for experimentation, team use and production. Use data policies to determine which links are permitted with which data.
• Get your data and rights in order. AI identifies everything a user has access to. Sensitivity labels and correct access rights are therefore not a mere formality, but a prerequisite.
• Keep track of AI usage outside your platform. With a solution such as Managed AI CloudGuard You can see which AI tools staff are using and decide what data can be sent to them.
• Make ownership clear. Every solution is assigned an owner. If a tool becomes business-critical, there is a clear path to professional management.
• Invest in your creators. Training and a dedicated point of contact lead to better solutions and reduced risk.

The role of IT is changing

This also shifts the role of IT. From a gatekeeper who builds or blocks everything themselves, to an enabler who provides the platform, the rules and the support. The business builds; IT ensures it remains secure and manageable. Those who consciously embrace this role turn vibe coding into a catalyst rather than a risk.

Facilitating and managing

At VanRoey ICT Group, we help organisations strike that balance. With our AI Implementation Programme We guide you through every stage, from strategy and governance to training and roll-out. And with our managed services, we continue to help safeguard your security, data and AI usage, even after the first apps have been built.

Would you like to know how ready your organisation is for the AI Vibe Coder? Our AI experts would be happy to discuss this with you. Please leave your details below to arrange a no-obligation consultation.