GEN_AI

GEN_AI is a collection of JavaScript experiments and applications for building LLM-powered assistants. The repository currently contains three separate projects:

Project	Purpose	Main technologies
Chatbot_agent	Browser chat UI backed by an Express API	React, Vite, Express, Groq, Exa
Company_Chatbot_RAG	Resume/company knowledge-base question answering	LangChain, Google embeddings, Pinecone, Groq
invokeLLM	Terminal chatbot with optional live web search	Node.js, Groq, Tavily

These projects are intentionally independent. Install dependencies and configure environment variables from the directory of the project you want to run.

Prerequisites

• Node.js 18 or newer
• npm
• An API key for the providers used by the selected project
• A terminal and a local clone of this repository

Do not commit .env files or API keys. The subprojects already contain local environment files or ignore rules; use those as a starting point and keep secrets private.

Quick start

Web chatbot

Open two terminals:

cd Chatbot_agent/Server
npm install
PORT=4000 npm start

Then start the UI in a second terminal:

cd Chatbot_agent/Client
npm install
npm run dev

The client calls http://localhost:4000/api/chat, so the server must use port 4000 unless the client request URL is changed in Chatbot_agent/Client/src/utils/api.services.js.

Retrieval-augmented chatbot

cd Company_Chatbot_RAG
npm install
node ingest.js
node index.js

Run ingestion after changing the source PDF. The Pinecone index and Google embedding configuration must be available before either command can complete. See Company_Chatbot_RAG/README.md for required variables.

Terminal web-search chatbot

cd invokeLLM
npm install
node app.js

Type questions at the You: prompt. Type exit to stop the process.

Repository layout

GEN_AI/
├── Chatbot_agent/
│   ├── Client/              React/Vite frontend
│   └── Server/              Express API and Groq agent
├── Company_Chatbot_RAG/     PDF ingestion and retrieval QA
└── invokeLLM/               CLI tool-calling example

How the projects differ

• Chatbot_agent keeps a conversation per browser-generated userId. The server stores recent message history in an in-memory cache and can call Exa when the model requests current information.
• Company_Chatbot_RAG retrieves relevant chunks from a Pinecone vector index before asking Groq to answer from the retrieved context. It is grounded in the indexed PDF rather than general web search.
• invokeLLM keeps a conversation in the terminal process and exposes Tavily as a model tool for current web results.

Common troubleshooting

• Missing API key: confirm the .env file is in the project directory where the Node process starts, then restart the process.
• Client cannot reach the API: verify the server port matches the URL in Client/src/utils/api.services.js and check that POST /api/chat is available.
• RAG returns no useful context: run node ingest.js with the intended PDF and verify the Pinecone index name, API key, namespace, and embedding model configuration.
• Provider limits or authorization errors: check the provider dashboard and the exact key expected by the selected project.

Development status

The projects are learning/prototype applications. The server packages do not currently define automated tests, and the chat server stores conversation history only in process memory. A production deployment would need persistent conversation storage, authentication, request validation, rate limiting, structured error handling, and secret management.