Agentic development in the NET Genium platform

Agentic development is a way of building and maintaining NET Genium applications in which an AI agent works alongside the developer – a tool that not only advises, but independently reads and modifies source files, the application model, and program code. Unlike a mere chat assistant, the agent has access to real development environment tools and carries out specific changes under the developer's supervision.

A typical agent is Claude Code connected to the NET Genium platform through an MCP server. As a result, the agent knows the framework architecture, the platform conventions, and the customer's specific implementation, and is able to work within it purposefully.

How agentic development differs from working with AI models

Both areas are closely related, but address a different task:

  • AI models provide an external assistant with knowledge of the structure of a specific implementation so that it can explain it and help with orientation. However, the assistant does not work with the application itself.
  • Agentic development goes a step further – based on this knowledge, the agent actually creates and modifies the application: it sets up forms and view pages, writes scripts and program code, and applies the changes directly to the application model.

AI models thus form the knowledge base, while agentic development is the driving force that uses this knowledge to build and maintain working solutions.

The MCP server as a knowledge layer

The connection between the agent and the NET Genium platform is provided by an MCP server (Model Context Protocol). It continuously supplies the agent, always in its current form, with:

  • the NET Genium framework documentation – a description of UI components, functions, the data model, and the architecture,
  • the binding rules and conventions of the platform – how to correctly design applications and write scripts and program code,
  • ready-made workflows for typical tasks – setting up an application, external functions, web services, or scheduled tasks.

As a result, the agent does not work according to general habits, but always according to the current NET Genium standards. When the framework evolves or the recommended practices change, the agent receives the up-to-date knowledge automatically through the MCP server – without the need to transfer anything manually.

What the agent can do

Equipped with knowledge of both the framework and the specific implementation, the agent can in particular:

  • design and set up new applications, edit forms, and view pages,
  • add and modify UI components, database queries, and server-side scripts,
  • write custom functions in C# – external functions, web APIs, and supporting services,
  • apply changes directly to the application model and prepare them for upload back to NET Genium,
  • process the requirements analysis and propose a corresponding data structure and processes.

In doing so, the agent respects the existing project structure and the style of the existing code – it does not introduce its own conventions, but mirrors what is already established in the given implementation.

A controlled and secure approach

Agentic development does not replace the developer, but significantly speeds up their work. All changes:

  • arise transparently as modifications to specific files and to the model,
  • are reviewed by the developer before deployment,
  • are versioned in source control, so they can be traced and reverted at any time.

This preserves the long-term sustainability, security, and continuity of the solution on which the NET Genium platform is built.

Typical use cases

Agentic development is particularly useful for:

  • speeding up the development of new applications and extending existing ones,
  • routine and repetitive tasks – setting up forms, UI components, and scripts,
  • writing and modifying program code beyond the capabilities of the graphical designer,
  • maintaining large-scale implementations while adhering to consistent conventions,
  • onboarding new developers, whom the agent helps to navigate both the implementation and the framework.