deft ferry
I'm wondering how HA and external AI agents communicate together, primarily, how HA receives and executes the commands prepared by AI agent after intent and task is understood.
I've had the general idea, however I asked GPT for his own view on that. I go this response:
https://chatgpt.com/s/t_6897441d09c48191ab2afe6f5be56633
There is this code, acting as a response, which seems logical:
{
"speech": {
"plain": {
"speech": "OK, turning on the kitchen lights."
}
},
"response": {
"type": "action",
"actions": [
{
"domain": "light",
"service": "turn_on",
"target": { "entity_id": ["light.kitchen"] }
}
]
}
}
- How can one debug the commands received from GPT (above JSON respone) for the given task requested in the AI assistant in the HA, like for example from the attached picture?
- How do we know what context has been provided to HA on the call so that it can understand the system setup?
