AI Processing Workflow

Full English transcript

Automatic English captions from the published YouTube video, lightly normalized for product names.

[0:00] To convert this project from KNX to Control4, we click here transform to Control4 and going to do is send the information about the structures of of the project and use AI to assist you to convert the actual structure of KNX to something more similar of what we want in Control4. For example, let me close a little bit here. Here you can see for example that I use prefix to order the rooms in KNX because KNX doesn't allow you to order the rooms. in this case the AI had sorry has removed the prefix. So also make a little bit more tidy namings. for example, living room have removed the the dash. and

[1:00] this give you the ability also to improve the matching of what is going to be be done about the AI. for example there there are situations where group addresses use other type of naming for example that the real rooms for example and master bedroom sometimes the group address has like something like primary bedroom. So the idea of this part is give you the the advantage of renaming things that are already there to be able to make a better match with AI but also we can very easily add buildings, floors and rooms inside what we have and we can rename for suppose that this is pool pool house.

[2:00] Sorry, old house. Then we have main and then we have for example sona and so on. We can add more room suppose that we have it in the group address. this way we are sending this information to the AI and we give them more context are more likely to find the devices and do the right mapping. So once you have validated this one important thing is you can't delete at this point rooms or readings on floors if they are from KNX and another thing I will recommend if you can tweak a little bit the names to make better match of the group addresses do it it's going to improve a lot and the mapping. So when it's done you go to next and here before processing the

[3:00] devices with the AI we have two prompt text one is general user prompt that can be configured in the settings of your user and you can enable if you want to use it or not in every project. this is the case for example you have always the same way to program or to do certain things. So you can use an general user prompt and use it when you want in every project. And then we have the project prompt that basically here we can give more information and more context to the AI. So we can improve the likelihood of matching. In the case of that the project is a little bit confusing once that we finish doing what we want. In this case I'm not going to put any prompt.

[4:00] we go to process device with AI and this is where all the information is sent so to to process and now we we must wait until the response. At this point you can still play with the platform the same if you go to the projects tab you can see that the is processing and the laps time and you can see that is process with AI. Right.