@Murat Zhandaurov I'm not sure I understand... You're trying to make a call directly from GHL? I don't think that is possible... There is something called call connect though

@Murat Zhandaurov I don't think the ghl node has that option, but you most likely could use an http node to trigger that functionality in ghl. I've never done that though. All my experience with voice agents is using vapi or retell.

I am not sure if I understand the question, but I think what you are saying is that when you test the solution, it works but when the client deploys the solution it does not work. Is that correct? If so we will need ALOT more information to help. maybe share your workflow and the error you are getting and what the flow is trying to achieve...

@Moses Nueman you could use a switch node to determine which route to take based on the user ID and have a dedicated branch for each user... this can obviously get messy the more and more users you have though

is it that LLMs arent allowed or that those specific companies arent allowed? as in, what about an on premises solutions, like hosting your own LLM? Not sure the answer to your question, also curious to see what others think.

There is a custom gpt for this already... I've only used it to see if it worked with a very simple workflow and it worked... Not sure how effective it is once the complexity of workflows increases but it is indeed possible to some degree... You could also build your own by scraping the n8n docs and the community node template and putting that into a rag system or maybe even training your own gpt4o via the openAI fine tune function... Again not sure how effective it would be but certainly possible... I'm thinking that it would be good enough to at least get a place to start a workflow, to get the bones of it, so to speak, but I'm thinking there will also be alot of fleshing out that will need to take place as well

I think the gpt is called n8n Superuser or something to that effect