Ice Breaker App

Overview

"The Glacier App" is a conversational helper tool that allows users to generate personalized conversation starters, summaries, and interesting facts about a person based on publicly available data from LinkedIn and Twitter. It leverages AI-based natural language processing with LangChain and APIs to gather data and present it in a user-friendly and engaging format.

This app can be particularly useful for networking professionals, recruiters, or anyone looking to make impactful initial impressions while engaging with someone new.

---

Features

• Smart Data Aggregation: Scrapes relevant information from LinkedIn and Twitter using intelligent agents.
• AI-Powered Insights: Generates an appealing summary of the person, unique facts, conversation starters, and topics of interest using AI.
• User-Friendly Web Interface: A simple, responsive front-end interface to enter names and view results quickly.
• Dynamic Spinner: Loading spinner to enhance the user experience while processing requests.
• Profile Picture Integration: Automatically fetches and displays the LinkedIn profile picture, if available.

---

How It Works

1. Input: Enter the name of the person whose insights you want to generate on the form.
2. Data Fetching:
   • Fetch LinkedIn and Twitter profile details using SerpAPIWrapper and connected APIs.
   • Scrape key information such as work history, tweets, topics of interest, and profile images.
3. AI-Powered Processing: Use the power of the LangChain library to:
   • Analyze the fetched data.
   • Generate conversational icebreakers and facts.
4. Output: View a summary, facts, ice-breakers, and suggested topics of interest alongside the profile image, all displayed clearly in the web interface.

---

Setup & Installation

Prerequisites

• Python 3.13.2 or higher
• pip for dependency management
• API keys for:
  • LinkedIn scraping
  • Twitter API
  • SerpAPI for Google Custom Search

Installation

1. Clone the repository:

git clone <repo_url>
   cd ice-breaker

2. Install dependencies:

pipenv install
   pipenv shell

3. Set up your .env file with the required API keys:

SCRAPIN_API_KEY=<Your_LinkedIn_API_Key>
   OPENAI_API_KEY=<Your_OpenAI_API_Key>
   SCRAPIN_API_KEY=<Your_Scripin_API_Key>
   TAVILY_API_KEY=<Your_Tavily_API_Key>
   LANGSMITH_API_KEY=<Your_Langsmith_API_Key>

4. Run the Flask application:

python app.py

5. Open your browser and navigate to:

http://127.0.0.1:5000/

---

Project Structure

• app.py: Main application file to run the Flask server and route API calls.
• templates:
  • index.html: The web interface used to input names and render the results.
• static/css:
  • style.css: Contains styles for the front-end interface.
• third_parties:
  • linkedin.py and twitter.py for data scraping from LinkedIn and Twitter respectively.
• agents: Lookup agents for fetching profile URLs based on name.
• glacier.py: Core business logic for creating conversational summaries and facts using LangChain.
• output_parsers.py: Defines output schemas for AI-generated data.
• tools.py: Custom SerpAPI wrapper for profile URL searches.

---

API Usage

/process [POST]

• Description: Processes the name submitted via the web form and returns the generated data.
• Request: { "name": "Harrison Chase" }
• Response:

{
    "summary": "Software Engineer focused on AI development.",
    "interests": ["Python", "Natural Language Processing"],
    "facts": ["Runs LangChain", "Has a passion for AI research"],
    "ice_breakers": ["What inspired your interest in AI?", "Do you teach others about LangChain?"],
    "picture_url": "https://linkedin.com/path/to/profile-pic.jpg"
  }

---

Technologies Used

Backend:

• Python 3.13.2
• Flask
• LangChain (for AI task orchestration)
• Tweepy (Twitter scraping)

Frontend:

• HTML/CSS
• FontAwesome (icons)

External APIs:

• ScrapIn (for linkedin profile lookups)
• Tavily API for scraping accessible data.