AI models in the NET Genium platform
The NET Genium platform allows creating and exporting AI models as a structured description of a specific application implementation. The model is derived from the platform's unified application model and describes applications, forms, view pages, relationships, rules, and system configuration.
In this concept, the AI model does not contain operational or user data. It serves as a safe knowledge base for specialized AI assistants in external tools (e.g. Claude, ChatGPT, Copilot Studio, etc.) that need to understand the structure of the solution, not the contents of stored records.
Purpose of the AI model
An AI model exported from NET Genium is intended to enable an external AI assistant to:
- understand the structure of a specific implementation
- know application groups, applications, and forms
- understand relationships between application building blocks
- explain the architecture of the solution
- assist with orientation in the implementation and documentation
The AI model is not intended for working with operational data or for querying the contents of individual forms.
Scope of information contained in the AI model
An exported AI model may include in particular:
- a list of application groups
- a list of applications
- edit forms and their structure
- view pages
- relationships between forms and applications
- basic descriptions of the purpose of individual parts of the system
- rules and constraints resulting from the implementation model
The model does not contain any user data, records, or values stored in the database.
Limits of the AI model
An AI assistant based only on this model:
- cannot answer queries such as "how many records exist"
- cannot search for specific items or documents
- cannot work with form contents
- has no access to live or historical data
Its knowledge is strictly limited to the structure and description of the system, not its data.
Creating an AI assistant from the model
Creating an AI assistant is always a manual process and consists of two basic steps:
Downloading the AI model from NET Genium
First, an AI model must be downloaded from a specific NET Genium implementation to a local environment. This model:
- corresponds to the current state of the application
- describes its structure and configuration
- serves as the knowledge base for the AI assistant
Creating the AI assistant in an external tool
The downloaded model is then manually used when creating an AI assistant in an external platform, for example:
- Claude
- ChatGPT
- Microsoft Copilot Studio
- other tools supporting custom models or instructions
NET Genium itself does not run or host the AI assistant – it only provides the underlying model.
Updating the AI model
When the application structure changes (e.g. form modification, adding an application, changing relationships), it is necessary to:
- download the current AI model to a local environment
- process the changes against the previous version
- upload the updated model back to the application
- manually update the AI assistant in the external tool
The AI model always reflects the state of the application at the time of the last upload.
Typical use cases
AI models in NET Genium are typically used for:
- orientation in complex implementations
- explaining system architecture
- onboarding new developers or administrators
- working with documentation
- consulting on application design and extension
- preparing the knowledge base for subsequent agentic development