Using ChatGPT for AVEVA Intouch, System Plaftorm and OMI

Hi Rainer,

As you can imagine we've also experimented with ChatGPT quite a bit  .

I am definitely interested in getting some more formal solutions, but I've had great success with scripting in particular. I just taught ChatGPT as best I could on how to write QuickScript.NET, and the results were pretty solid if not perfect. I learned early on to tell it to only support .NET 1.1 to avoid it trying to use things like generics and lists etc. I also splurged on the paid-for version of ChatGPT, which allowed me to feed it the Scripting.PDF document. One of the early successes I had was when I wanted to integrate weather information into objects in System Platform. ChatGPT did not only write the script for me, but helped me get access to the correct (free ) website that would supply the necessary information for me. At the time it took me less than 10 minutes to implement and I was really impressed!

One of the things that we want to improve in an upcoming version is to allow more text-based exports of our configuration. This would include an XML-based export/import of Industrial Graphics (you can already do this today using the GR Access toolkit, but I want this to be a menu option). One of the things that this would allow us to do is to teach ChatGPT what the inside of that XML looks like and of course how to create one on its own... I believe that would lay the groundwork for a GPT that would allow us to create industrial graphics!

We are also working on exporting of other configurations in text-based formats including Templates and Attributes which should further allow us to teach ChatGPT how to create a model and probably more importantly the templates used to create that model.

Up until that point we would still be using ChatGPT in its native format which as you know can sometimes result in hours of experimentation to attempt to find the ultimate prompt . But once all of these things are in place, I think it would just be a short step to create a custom GPT that we can teach up front to do a lot of these things. One of the challenges may be to find examples to feed it but I am sure that if we could overcome this, we could build a special "System Platform GPT" that can be a very quick leg-up when creating a new solution.

Of course, I am looking at this through my "System Platform & OMI" eyes and other products may see other opportunities to leverage it.

Hi Rainer,

As you can imagine we've also experimented with ChatGPT quite a bit  .

I am definitely interested in getting some more formal solutions, but I've had great success with scripting in particular. I just taught ChatGPT as best I could on how to write QuickScript.NET, and the results were pretty solid if not perfect. I learned early on to tell it to only support .NET 1.1 to avoid it trying to use things like generics and lists etc. I also splurged on the paid-for version of ChatGPT, which allowed me to feed it the Scripting.PDF document. One of the early successes I had was when I wanted to integrate weather information into objects in System Platform. ChatGPT did not only write the script for me, but helped me get access to the correct (free ) website that would supply the necessary information for me. At the time it took me less than 10 minutes to implement and I was really impressed!

One of the things that we want to improve in an upcoming version is to allow more text-based exports of our configuration. This would include an XML-based export/import of Industrial Graphics (you can already do this today using the GR Access toolkit, but I want this to be a menu option). One of the things that this would allow us to do is to teach ChatGPT what the inside of that XML looks like and of course how to create one on its own... I believe that would lay the groundwork for a GPT that would allow us to create industrial graphics!

We are also working on exporting of other configurations in text-based formats including Templates and Attributes which should further allow us to teach ChatGPT how to create a model and probably more importantly the templates used to create that model.

Up until that point we would still be using ChatGPT in its native format which as you know can sometimes result in hours of experimentation to attempt to find the ultimate prompt . But once all of these things are in place, I think it would just be a short step to create a custom GPT that we can teach up front to do a lot of these things. One of the challenges may be to find examples to feed it but I am sure that if we could overcome this, we could build a special "System Platform GPT" that can be a very quick leg-up when creating a new solution.

Of course, I am looking at this through my "System Platform & OMI" eyes and other products may see other opportunities to leverage it.

Hi Ernst, this is what we are looking for, adding any data to make our developer lives easy. Your example for weather data is exactly what we will be using ChatGPT for.  Bang on on the export and import as an easy option, thats exactly what we need. 

Honestly we would like to use ChatGPT in its native format rather than a product, maybe you can have an add on within ChatGPT if you need to incentivise on it , riding the wave of what you suggested, i would like some more things:

1) Creation of tags as a JSON file that i can import within AVEVA , if chatgpt can create this JSON from excel input, that would be amazing

2) Simple xml to change some graphics

3) Scripting ( but you already touched on it )

4) Understanding logs, say if put the logs from OCMC can it tell me what happened

I think even if we have some videos on how to use ChatGPT in its native format with Intouch/ AVEVA Portfolio that would be amazing, product intergation can come much later