Leadgen AI is a powerful tool that lets you extract data from any website URL with ease.
Started with a script using Selenium (for dynamic pages) and Beautiful Soup to scrape contact info (emails, phone numbers) and social media links from any URL.
Cleaned and formatted the scraped data, splitting it into chunks to stay within LLM token limits.
Chose Llama 3.1 for its speed and cost-efficiency. It analyzes scraped content and generates personalized outreach (cold emails, LinkedIn requests, follow-ups) based on user prompts.
Built a FastAPI backend and connected it to Google Sheets via Apps Script for seamless usage.
Integrated a chat sidebar where users can select templates or enter custom prompts—Llama 3.1 handles the messaging instantly.
Scraping Accuracy: ~90% (manual testing)
Speed: 2–5s for scraping, 7–10s for responses
Impact: Cuts outreach drafting time by ~70%
Install the required Python libraries:
pip install -r requirements.txt-
Open your Google Sheet
-
Go to Extensions → Apps Script
-
Open the google-sheets directory from the repo
-
Copy and paste the script into the Apps Script editor and save
Run the FastAPI server locally:
uvicorn main:app --reloadMake your local server publicly accessible:
ngrok http 8000Paste URLs into your sheet. Then use the custom formula to extract data:
=EXTRACT_DATA(urlrange, fieldrange)
Example:
=EXTRACT_DATA(A2:A4, B1:E1)
This will scrape the URLs in A2:A4 and return the fields specified in B1:E1.
Successfully extracted data from Ycombinator's public page
Generating an outreach mail for collaboration with reddit
Output of the outreach mail with the prompt "We are an upcoming social media platform called leadgen"
