Greetings Community,
I'm really concerned about the chatbot I'm working on. I've encountered several difficulties. I've been working for over 3 months with Copilot Studio, and they're on the verge of canceling my project due to the errors encountered. My client is upset.
1.) I have around 80 questions, and some questions don't match the corresponding topic.
2.) We've conducted several tests with real people asking questions, but it still feels very robotic. (Is there a way to make it more user-friendly without needing to add the question and answer?)
Example: Thank you responses, ongoing conversation, virtual assistant.
3.) Some questions aren't trained properly, and even though I've added them, it doesn't select the correct topic.
4.) Too many errors have occurred for the assistant to understand the question and provide a suitable response.
Is there a way to connect it to ChatGPT and have it automatically generate conversational responses as if it were a real person? Also, can it generate topic information when detected?
We've faced significant challenges because a user can have different communication styles.
