🚀 NexMind — An Extensible AI Workbench with Pluggable Agents (Ollama + Gemini)

🚀 NexMind — An Extensible AI Workbench with Pluggable Agents (Ollama + Gemini)

Developers use AI tools for many small tasks every day:

chatting with LLMs

translating text

summarizing documents

testing prompts

experimenting with RAG

But most of the time we end up using multiple different tools.

So I decided to build something to simplify this workflow.

Introducing NexMind — an extensible AI workbench where developers can plug in different AI agents and switch models instantly.

🔗 GitHub Repo:
https://github.com/Pandi2352/NexMind

🧠 What is NexMind?

NexMind is a developer-focused AI workbench that allows you to:

configure multiple AI providers

assign providers per AI agent

integrate vector databases for RAG

switch models without restarting

experiment with AI workflows in one place

It currently supports:

Ollama (local or remote models)

Google Gemini

The system is built so new AI agents can be added easily.

🏗 Architecture

The project is split into two main parts:

nexmind/
├── server/ → NestJS REST API
└── web/ → React + Vite frontend
Backend

NestJS

MongoDB

LangChain

Swagger API docs

Frontend

React 19

TypeScript

Tailwind CSS

React Router

Axios

The architecture is designed to be provider-agnostic, meaning you can switch LLM providers without changing agent logic.

🔌 AI Provider Management

NexMind lets you configure multiple AI providers.

Supported providers:

Ollama (local)

Ollama cloud

Google Gemini

You can:

✔ Add multiple providers
✔ Set an active provider
✔ Assign specific providers per agent

Example setup:

Chat Agent → Gemini
Translator → Ollama
RAG Chat → Ollama + Vector Store

No server restart required.

🧩 Built-in AI Agents

NexMind includes several built-in agents.

💬 Chat

Multi-conversation chatbot with message history and system prompts.

📚 RAG Chat

Chat augmented with retrieval from vector databases.

🌍 Translator

AI-powered text translation with language detection and history.

✂️ Summarizer

Summarize text using styles like:

Bullet points

TL;DR

ELI5

🧠 Prompt Optimizer

Transforms vague prompts into clear structured instructions.

🩺 Health Advisor

Experimental AI assistant for:

symptom analysis

diet suggestions

wellness guidance

🗄 Vector Store Support

For RAG capabilities, NexMind supports:

Pinecone

Chroma

Milvus

Qdrant

Upstash

Each agent can be configured with its own vector store.

⚙️ Example Workflow

Example configuration:

Provider: Gemini → Chat
Provider: Ollama → Summarizer
Vector Store: Pinecone → RAG Chat

This allows each agent to use the best provider for its specific task.

🚀 Quick Start

Clone the repository:

git clone https://github.com/Pandi2352/NexMind

Start backend:

cd server
npm install
npm run start:dev

Start frontend:

cd web
npm install
npm run dev

Open:

http://localhost:5173

From there you can:

add AI providers

configure vector stores

assign agents

start chatting with AI

📦 Tech Stack

Backend

NestJS

MongoDB

LangChain

Swagger

Frontend

React 19

TypeScript

Tailwind CSS

Vite

🔮 Future Plans

Some features I plan to add:

plugin system for custom agents

document ingestion pipelines

streaming responses

OpenAI / Anthropic provider support

workflow automation

💡 Feedback Welcome

This project is still evolving, and I’d love feedback from other developers.

What agents would you like to see next?

⭐ If you find it interesting, feel free to check out the repo and leave a star:

https://github.com/Pandi2352/NexMind