Prompt response quality in Conversational Agent system message

[natellanur]

Hi all,

I have a prompt that I have tested multiple times with OpenAI directly and via API calls from bubble.io and it works well there. But when I try to use it with Flowise - it works terrible. The chat doesn't do what I ask it to do, misunderstands me, or ignores some parts. In Flowise I tested it with Conversational Agent system message and am wondering what could be the reason for this behavior.

Will appreciate some thoughts on this.