app.py

import gradio as gr
from transformers import pipeline
import crewai

# Load an open-source reasoning model
conversation_model = pipeline("text-classification", model="tiiuae/falcon-7b-instruct")

# Define conversation steps
STEPS = [
    "Salutations",
    "Welcome message",
    "Validating customer account",
    "Asking for problem",
    "Working on issue",
    "Escalating if not able to handle",
    "Informing customer about solutions",
    "Asking customers if they have any other issue",
    "Saying 'Have a great day'"
]

def detect_steps(transcript):
    detected_steps = []
    
    for step in STEPS:
        result = conversation_model(transcript, return_all_scores=True)
        
        # Simple heuristic: If model confidence is above 0.5 for the step, mark it as completed
        step_detected = any(label["label"].lower() in step.lower() and label["score"] > 0.5 for label in result[0])
        
        if step_detected:
            detected_steps.append(f"✔️ {step}")
        else:
            detected_steps.append(f"❌ {step}")
    
    return "\n".join(detected_steps)

# Gradio interface
demo = gr.Interface(
    fn=detect_steps,
    inputs=gr.Textbox(placeholder="Paste live transcript here..."),
    outputs=gr.Textbox(label="Detected Steps"),
    title="Live Conversation Step Detector",
    description="Paste a live transcript of a conversation between an Agent and a Customer. The AI will detect and mark completed steps."
)

if __name__ == "__main__":
    demo.launch()