Neele-Vat is a family-owned logistics company with more than 1,700 employees, headquartered in Rotterdam. It manages complex global supply chains, including the specialized transport of hazardous goods. Its core customer application, which handles shipping and tracking, originally ran on a dedicated Virtual Machine (VM) setup. With mogenius for the infrastructure and Tetrifox for software and consulting, Neele-Vat first moved that application to a production-grade Kubernetes platform on Microsoft Azure. It then used the same platform to get ahead of a second shift: generative AI putting software creation into the hands of far more people than the development team alone.
Before the migration, Neele-Vat's development and operations teams worked against an inefficient setup. Test and production ran on the same VM, which created instability and risk, and every update depended on one person manually managing that machine. That made the individual a bottleneck and stalled feature development.
The team was strong in application development but had no in-house Kubernetes specialists. They understood the value of isolated environments and modern tooling, but they were application developers, not platform engineers, and the goal was to empower them rather than turn them into Kubernetes experts overnight. The single-VM architecture also could not handle production loads for I/O-heavy components such as EventStoreDB, and it could not support the company's growth plans or its planned AI initiatives.
“The operational complexity of Kubernetes was a significant hurdle for our team. mogenius made it simple and manageable for all our developers to adopt and use.”
Neele-Vat already worked with Tetrifox. For the move to Kubernetes it needed a second partner, not just a product: someone to build an entire Kubernetes platform, from the cluster up to developer tooling. mogenius was chosen for its full-spectrum professional services and the abstraction layer of the mogenius Kubernetes platform, which was decisive for a team without deep Kubernetes expertise.
From there, mogenius provided the platform and the infrastructure expertise, while Tetrifox stayed on the software side. Together this covers the whole path, from the cluster to the applications running on it.
The rollout was fast and low-risk. The mogenius operator was provisioned directly with the new Azure Kubernetes Service (AKS) clusters, so the setup stayed transparent to the Neele-Vat team.
Since the migration, AI has changed the picture inside most organizations, and Neele-Vat saw it coming early. Generative AI has become an everyday tool far beyond the software team. People in every department can now describe what they need and have tools like Claude generate a working container application for them. That is a real productivity gain, and it also creates a new class of risk that most companies only notice once it is already widespread.
A year ago, keeping software standardized meant governing perhaps 20 developers. In the age of AI, the group that can produce a deployable app is no longer 5 to 10 percent of staff, in principle it is every employee, from engineering to the front desk. The scale of who can create software has changed by an order of magnitude.
The fastest way for a non-technical builder to get an app running is a Platform-as-a-Service shortcut, where hosting is fully abstracted away. That convenience is exactly the problem for an enterprise: those environments sit outside IT's line of sight, with no governance, no security checks, and no control over where data is processed. Enterprise-grade protections do not apply to apps the organization never sees. For a logistics company with strict data-protection obligations, that is a gap worth closing before it opens. Rather than wait for that to happen, Neele-Vat used its mogenius platform to put a governed path in place ahead of demand, so that as employees began building with AI, there was already a compliant home for those apps.
Neele-Vat's answer was not to forbid the behavior but to make the safe path the easy path. The organization now offers its people a way to run their AI-built apps, internally called citizen apps, on the same central Kubernetes platform, with enterprise guardrails built in.
The result is a path where a person with no software background can take the container app their AI built and run it on Kubernetes: a good user experience, no shadow IT, and the central enterprise offering instead. The app runs securely, auditably, and within governance.
The honest lesson from this use case, and one that applies to any organization, is that you cannot simply forbid AI-driven building. Policies that ban it are already in place at many companies, and people work around them anyway. You can steer behavior only by offering an alternative with the same convenience, pointed in the right direction.
In practice that means a standard process, communicated clearly: if you want to deploy an AI-built app, it runs through mogenius. Access is requested with an existing login, the mogenius MCP server can be distributed centrally as a connector, and the documentation lives in one place. Organizations can reinforce this by blocking external Platform-as-a-Service domains, so that when a builder asks their agent to deploy externally, the agent responds that the company does not allow it and offers the governed mogenius path instead. You cannot remove the choice, but you can make the governed path the obvious one.
“We did not want to slow the teams down. We wanted the easy way and the safe way to be the same way. mogenius let Neele-Vat give everyone a governed path to run what they build.”
The Neele-Vat platform was built once and now does two jobs: it runs the company's core logistics applications on modern Kubernetes, and it gives every AI-assisted builder a governed place to ship. The wider takeaway is the forward-looking one: the wave of AI-built software is coming to every organization whether IT is ready or not, and the teams that prepare a compliant, self-service path in advance are the ones who turn that wave into an advantage instead of a liability.
That is usually the point where a governed, self-service path for citizen apps is worth a conversation. Tell us about your setup, or read how other teams did it.
Tetrifox helps businesses scale. Our teams become part of the client’s organisation, combining business understanding, technical leadership and engineering. We take responsibility from the first idea through to production and ongoing operation. Software is the tool; the goal is a business that can grow without technology holding it back.
The core application ran on a dedicated VM with no CI/CD, which forced developers to rely on a single person for manual deployments. That lack of automation and isolation brought development of new features to a halt, and the single VM could not scale for growth or for the company's planned AI initiatives.
Neele-Vat needed a strategic partner offering full-spectrum professional services to build the entire platform, not just a product. That bridged the internal skill gap and gave developers an intuitive platform to adopt quickly, without becoming infrastructure specialists.
The Kubernetes infrastructure, covering Development, UAT, and Production on Azure Kubernetes Service, was established in about one month, using an Infrastructure as Code approach with Bicep.
Shadow IT is software that runs outside an organization's governance. AI tools make it far more common, because employees with no development background can now generate and deploy working apps. The group that can create software has grown from a small number of developers to potentially every employee, which multiplies the ungoverned surface.
Builders select simple templates through a self-service interface, or ask an AI agent to deploy for them through the MCP server. Each template carries network policies, Kyverno policies, and namespace isolation under the hood, so the right security configuration is applied automatically without the builder needing Kubernetes knowledge.
Every action, whether from the UI, Git, the API, or an AI agent through MCP, resolves to the same platform with the same audit trail. Agent actions are attributed to the developer they act on behalf of, so deployments made in natural language remain fully traceable.
Subscribe to our newsletter and stay on top of the latest developments
