atvise® & AI
Why the Foundation Determines the Benefits
AI needs structure. atvise® has had it since 2006.
The answer is already in the project. You must be able to access it.
atvise® and AI: Why open standards determine whether artificial intelligence delivers value in automation - and what the team is currently working on.
| An example: - A plant reports a fault. - The responsible technician is not on site. - The colleague filling in does not know the project. - He does not know why an alarm is configured the way it is. - Which script is triggered where? - But: the information is already in the system! - It is stored in the project, the documentation and the log files. - It is simply not accessible at the moment it is needed... |
This is exactly where artificial intelligence in automation becomes interesting - not as a replacement for subject matter expertise, but as a path toward it. And this is exactly where it becomes clear that the crucial question is not the model, but the system underneath it.
An assistant is only as good as its access to the system
A language model does not know the system. It can only answer what it can read. In a closed system this means: It reads very little and what it does read must be provided via custom-built intermediate layers. Each of these layers represents effort, a source of error, and a dependency.
With atvise® this detour is eliminated. The foundation for this was laid long before the current AI discussion.
Three Essentials for AI: atvise® has them.
For an assistant to work with a plant, it needs more than a language model. It needs structured access to information, an environment for application, and open interfaces to the outside world. These foundations have been part of atvise® for years.
The foundation for AI has been in place for years:
- OPC UA structures the information.
- Web technology makes it accessible everywhere.
- Open interfaces, scripts, and OPC UA methods provide the path to the outside world.
What AI applications need today is therefore not an entirely new technological world. They build on an architecture that has shaped atvise® for years.
Current Status
The atvise® architecture has been systematically analyzed to determine where AI provides real value and where it does not. The first assistive tools have been developed and are currently in trial operation:
- tools to assisst Dashboard configuration
- Documentation
- Application-related questions
The greatest potential lies not in runtime operation, but in engineering. This is where a significant amount of effort is generated that can be reduced substantially with AI assistance. It is also where errors can be identified and corrected before they affect the plant. In the engineering role, agents can support tasks such as generating scripts, displays, or object types.
In runtime, agents play a different role, depending on their tasks and the way they access the process image:
Reading | non critical | Tools and agents can read information. |
Writing | Critical | Process interventions and write operations require control. |
In automation a misspelled variable can alter the set value, destroy a display, or override an alarm configuration.
That is why the principle is adhered to:
- Every intervention by an assistant requires explicit human approval.
- Every suggestion must be reviewed.
What comes next...
...the foundation is in place.
We will continue this work over the coming months and show what emerges from it.
- You’ll be the first to know, right here.
More Technologies
Built for Innovation
