Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
from langchain_community.document_loaders import AsyncHtmlLoader
|
| 3 |
+
from langchain_community.document_transformers import Html2TextTransformer
|
| 4 |
+
import tiktoken
|
| 5 |
+
from langchain_groq import ChatGroq
|
| 6 |
+
import streamlit as st
|
| 7 |
+
from dotenv import load_dotenv
|
| 8 |
+
from pathlib import Path
|
| 9 |
+
|
| 10 |
+
env_path = Path('.') / '.env'
|
| 11 |
+
load_dotenv(dotenv_path=env_path)
|
| 12 |
+
|
| 13 |
+
st.title("AI Sales Executive")
|
| 14 |
+
urls_input = st.text_area("Enter website URLs (comma-separated):")
|
| 15 |
+
|
| 16 |
+
if st.button("Submit"):
|
| 17 |
+
if urls_input:
|
| 18 |
+
urls = [url.strip() for url in urls_input.split(",")]
|
| 19 |
+
loader = AsyncHtmlLoader(urls)
|
| 20 |
+
docs = loader.load()
|
| 21 |
+
|
| 22 |
+
html2text = Html2TextTransformer()
|
| 23 |
+
docs_transformed = html2text.transform_documents(docs)
|
| 24 |
+
|
| 25 |
+
llm = ChatGroq(
|
| 26 |
+
model="llama3-8b-8192",
|
| 27 |
+
temperature=0,
|
| 28 |
+
max_tokens=None,
|
| 29 |
+
timeout=None,
|
| 30 |
+
max_retries=2,
|
| 31 |
+
)
|
| 32 |
+
|
| 33 |
+
enc = tiktoken.encoding_for_model("gpt-3.5-turbo")
|
| 34 |
+
|
| 35 |
+
prompt = """You are a senior sales executive tasked with demonstrating how your expert team of data scientists can significantly enhance this company's growth and optimize their existing products using AI/ML technologies. Provide detailed insights into the specific ways your team can contribute to the company's success, specifically tailored to the company's product and goals. Additionally, include a brief summary of the company based on the following website content:
|
| 36 |
+
|
| 37 |
+
Website content: {content}
|
| 38 |
+
"""
|
| 39 |
+
content = """"""
|
| 40 |
+
for doc in docs_transformed:
|
| 41 |
+
content += doc.page_content + "\n\n"
|
| 42 |
+
|
| 43 |
+
with st.spinner("Generating response..."):
|
| 44 |
+
response = llm.invoke(prompt.format(content=content))
|
| 45 |
+
st.write(response.content)
|