Examples Repository

This repository contains a collection of examples demonstrating the usage of LangGraph in various applications. LangGraph is a powerful framework for building dynamic, decision-making workflows with natural language processing capabilities.

---

Overview

LangGraph provides a way to model complex conversational flows, manage state, and integrate external tools and APIs seamlessly. The examples in this repository aim to showcase the versatility of LangGraph for different real-world use cases, from personal assistants to privacy-focused agents.

Each folder represents a different project or example that demonstrates how LangGraph can be combined with various technologies like OpenAI models, audio-based input/output, and external web services.

---

Featured Examples

1. Voice Agent
   This project demonstrates a voice-based agent that uses:

   • Llama 3.2 (3B) for natural language understanding (via Ollama)
   • Whisper for local audio transcription
   • Kokoro for local text-to-speech (TTS)
   • DuckDuckGo for privacy-focused web searches.

   Note: This project runs 100% locally, using open-source models to ensure privacy and control over data.
   Link to Voice Agent Project

2. Vision Agent
   This project showcases a vision-based agent capable of analyzing both text and images. It uses:

   • Llama 3.2 Vision (11B) via Ollama for multimodal understanding.
   • LangGraph to dynamically handle workflows and integrate visual and textual data.
   • Built-in image preprocessing, where images are converted to Base64 strings for seamless compatibility.
   • Advanced vision capabilities, including:
     • Object detection and scene understanding.
     • OCR for extracting text from images.
     • Multimodal question answering combining text and visual inputs.

   Note: This project ensures privacy by running all computations locally and leveraging cutting-edge models for visual analysis.
   Link to Vision Agent Project

3. Web Browsing Agent
   This project highlights a web browsing agent capable of navigating websites based on user input. It uses:

   • Llama 3.2 (via Ollama) for natural language understanding and decision-making.
   • Helium for lighter web automation with Python.
   • Selenium for browser automation.

   The agent can:

   • Open a web browser and navigate to specified URLs.
   • Perform searches and interact with web elements dynamically.
   • Execute tasks entirely on a local machine, ensuring privacy and security.

   Note: This project runs 100% locally, leveraging powerful tools for automated web interaction.
   Link to Web Browsing Agent Project

---

Future Updates

This repository will be gradually updated with more examples showcasing the diverse applications of LangGraph. Stay tuned for new projects and enhancements!

---