AI Assistant

A 𝐩𝐫𝐢𝐯𝐚𝐜𝐲-𝐟𝐢𝐫𝐬𝐭 𝐑𝐀𝐆 𝐀𝐈 𝐚𝐬𝐬𝐢𝐬𝐭𝐚𝐧𝐭 𝐟𝐨𝐫 𝐖𝐨𝐫𝐝𝐏𝐫𝐞𝐬𝐬.
The RAG application runs with a 𝐩𝐫𝐢𝐯𝐚𝐭𝐞𝐥𝐲 𝐡𝐨𝐬𝐭𝐞𝐝 𝐋𝐋𝐌, ensuring your data never leaves your infrastructure - no third-party AI APIs, no external data sharing.

🧠 Dynamic data indexing — create, update, or delete your AI knowledge base directly from the WordPress admin panel
📚 RAG-powered, only context-aware answers
🔐 Full data ownership & protection
🧩 Native WordPress UI

Architecture Overview

┌──────────────────────────────────────────────────────────────────────┐
│                          User Interface                              │
│                     (WordPress Admin Panel)                          │
│                         http://localhost:8000                        │
│  Provides interactive web interface for document management          │
│  and AI-powered question answering                                   │
└────────────────────────────────┬─────────────────────────────────────┘
                                 │
                                 │ User Interaction
                                 ▼
┌──────────────────────────────────────────────────────────────────────┐
│              WordPress Plugin (assistant-interface)                  │
│         Location: assistant-interface/wordpress/wp-content/          │
│                      plugins/assistant-interface/                    │
│  ┌───────────────────────────────────────────────────────────────┐   │
│  │  Frontend Layer (React + Webpack)                             │   │
│  │  • Renders responsive UI components                           │   │
│  │  • Handles user input and form submissions                    │   │
│  │  • Manages client-side state and interactions                 │   │
│  │  • Compiles JSX/CSS into optimized bundles                    │   │
│  └───────────────────────────────────────────────────────────────┘   │
│  ┌───────────────────────────────────────────────────────────────┐   │
│  │  Backend Layer (PHP)                                          │   │
│  │  • Processes HTTP requests from frontend                      │   │
│  │  • Manages WordPress integration and hooks                    │   │
│  │  • Forwards API calls to semantic-engine                      │   │
│  │  • Handles authentication and authorization                   │   │
│  └───────────────────────────────────────────────────────────────┘   │
│  ┌───────────────────────────────────────────────────────────────┐   │
│  │  Infrastructure (Docker)                                      │   │
│  │  • Nginx serves static assets and routes requests             │   │
│  │  • PHP-FPM processes WordPress application logic              │   │
│  │  • MySQL stores WordPress data and configurations             │   │
│  └───────────────────────────────────────────────────────────────┘   │
└────────────────────────────────┬─────────────────────────────────────┘
                                 │
                                 │ REST API Calls (HTTP/HTTPS)
                                 │
                                 ▼
┌──────────────────────────────────────────────────────────────────────┐
│              RAG Backend (semantic-engine)                           │
│              Location: ai-assistant/semantic-engine/                 │
│  ┌───────────────────────────────────────────────────────────────┐   │
│  │  Indexing Module (app/index/)                                 │   │
│  │  • Dynamically indexes documents in vector database           │   │
│  │  • Processes and chunks documents for embedding               │   │
│  │  • Manages document lifecycle (store/delete/re-index)         │   │
│  │  • Generates vector embeddings using Ollama                   │   │
│  │                                                               │   │
│  │  API Endpoints:                                               │   │
│  │    POST   /index/store     - Index new documents              │   │
│  │    DELETE /index/delete    - Remove indexed documents         │   │
│  │    POST   /index/re-store  - Reindex existing documents       │   │
│  └───────────────────────────────────────────────────────────────┘   │
│  ┌───────────────────────────────────────────────────────────────┐   │
│  │  Retrieval Module (app/retrieval/)                            │   │
│  │  • Performs semantic search on user queries                   │   │
│  │  • Augments prompts with retrieved context                    │   │
│  │  • Generates AI responses using RAG approach                  │   │
│  │  • Retrieves relevant document chunks and metadata            │   │
│  │                                                               │   │
│  │  API Endpoints:                                               │   │
│  │    POST /retrieval          - Query with RAG                  │   │
│  │    GET  /retrieval/document - Fetch specific document         │   │
│  │    GET  /retrieval/documents- List all documents              │   │
│  │    GET  /retrieval/points   - Get vector points               │   │
│  └───────────────────────────────────────────────────────────────┘   │
│  ┌───────────────────────────────────────────────────────────────┐   │
│  │  Infrastructure (Docker)                                      │   │
│  │  • Node.js runtime executes application logic                 │   │
│  │  • Nginx reverse proxy with SSL termination                   │   │
│  │  • Ollama integration for embeddings and generation           │   │
│  └───────────────────────────────────────────────────────────────┘   │
└────────────────────────────────┬─────────────────────────────────────┘
                                 │
                                 │ Vector Operations
                                 │ (Embedding Storage & Retrieval)
                                 ▼
┌──────────────────────────────────────────────────────────────────────┐
│                  Vector Database (Qdrant)                            │
│                      Storage: qdrant_db/                             │
│  • Stores high-dimensional vector embeddings                         │
│  • Performs fast similarity search using HNSW algorithm              │
│  • Manages document metadata and payloads                            │
│  • Enables semantic retrieval for RAG operations                     │
└──────────────────────────────────────────────────────────────────────┘
                                 ▲
                                 │
                          Powered by Ollama
                                 │
                    ┌────────────────────────────┐
                    │    Ollama (Local AI)       │
                    │  • Generates embeddings    │
                    │  • Produces AI responses   │
                    │  • Runs locally on host    │
                    └────────────────────────────┘

Conversation flow

The RAG system maintains conversation context through an intelligent summarization approach that preserves conversation history while optimizing token usage.

Flow Diagram

┌─────────────────────────────────────────────────────────────────────┐
│                     First Question (Initial Query)                  │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  User Query                                                         │
│       │                                                             │
│       ├─► Semantic Search (Qdrant)                                  │
│       │                                                             │
│       ├─► Context Retrieval                                         │
│       │                                                             │
│       ├─► Prompt Augmentation                                       │
│       │                                                             │
│       └─► AI Response Generation                                    │
│                                                                     │
│  Output:                                                            │
│    • AI Answer                                                      │
│    • Summary (Initial Context)                                      │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────────────┐
│              Second and Nth Question (Context-Aware Query)          │
├─────────────────────────────────────────────────────────────────────┤
│                                                                     │
│  User Query + Previous Summary                                      │
│       │                                                             │
│       ├─► Semantic Search (with conversation context)               │
│       │                                                             │
│       ├─► Context Retrieval                                         │
│       │                                                             │
│       ├─► Prompt Augmentation                                       │
│       │   (Previous Summary + New Context)                          │
│       │                                                             │
│       └─► AI Response Generation                                    │
│                                                                     │
│  Output:                                                            │
│    • AI Answer                                                      │
│    • Updated Summary                                                │
│                                                                     │
└─────────────────────────────────────────────────────────────────────┘

Benefits of This Approach

Context Preservation: Each question builds upon previous interactions
Token Efficiency: Summaries compress conversation history without losing key information
Semantic Continuity: Follow-up questions can reference earlier topics naturally
Scalable Conversations: Long conversations remain manageable through progressive summarization
Enhanced Retrieval: Cumulative context improves relevance of retrieved documents

Project Structure

ai-assistant-services/
│
├── ai-assistant/                          # Main Git Repository
│   ├── semantic-engine/                   # RAG Backend Application
│   │   ├── app/
│   │   │   ├── index/                     # Document Indexing Module
│   │   │   │   ├── controllers/
│   │   │   │   │   ├── storeController.js
│   │   │   │   │   ├── deleteController.js
│   │   │   │   │   └── reStoreController.js
│   │   │   │   ├── services/
│   │   │   │   │   ├── createDocumentService.js
│   │   │   │   │   └── deleteDocumentService.js
│   │   │   │   └── routes.js
│   │   │   │
│   │   │   ├── retrieval/                 # Query & Retrieval Module
│   │   │   │   ├── controllers/
│   │   │   │   │   ├── retrievalController.js
│   │   │   │   │   ├── getDocumentController.js
│   │   │   │   │   ├── getDocumentsController.js
│   │   │   │   │   └── getPointsController.js
│   │   │   │   ├── services/
│   │   │   │   │   └── prompt-augmentation.js
│   │   │   │   └── routes.js
│   │   │   │
│   │   │   ├── helpers/
│   │   │   │   └── utilities.js
│   │   │   └── types/
│   │   │       └── point.js
│   │   │
│   │   ├── docker-configs/                # Docker Configuration
│   │   │   ├── certs/
│   │   │   │   ├── cert.pem
│   │   │   │   └── key.pem
│   │   │   ├── default.conf
│   │   │   ├── nginx.dockerfile
│   │   │   └── node.dockerfile
│   │   │
│   │   ├── qdrant_db/                     # Vector Database Storage
│   │   ├── app.js                         # Main Application Entry
│   │   ├── docker-compose.yml
│   │   ├── package.json
│   │   ├── nodemon.json
│   │   ├── .env
│   │   └── .env-example
│   │
│   └── assistant-interface/               # a copy of the plugin
│
└── assistant-interface/                   # WordPress Development Environment
    ├── wordpress/                         # WordPress Core
    │   ├── wp-content/
    │   │   ├── plugins/
    │   │   │   └── assistant-interface/   # Main WordPress Plugin
    │   │   │       ├── app/
    │   │   │       │   ├── Core.php       # Plugin Core
    │   │   │       │   ├── Lib/           # Library Classes
    │   │   │       │   │   ├── Response.php
    │   │   │       │   │   ├── Http.php
    │   │   │       │   │   └── SingleTon.php
    │   │   │       │   ├── Services/      # Business Logic
    │   │   │       │   │   ├── KnowledgeBase.php
    │   │   │       │   │   ├── AdminInterface.php
    │   │   │       │   │   └── Assets.php
    │   │   │       │   └── Views/
    │   │   │       │       └── admin.php
    │   │   │       │
    │   │   │       ├── resources/         # Frontend Assets (Source)
    │   │   │       │   ├── components/
    │   │   │       │   │   └── IndexForm.jsx
    │   │   │       │   ├── app.js
    │   │   │       │   └── app.css
    │   │   │       │
    │   │   │       ├── build/             # Compiled Frontend Assets
    │   │   │       ├── vendor/            # PHP Dependencies (Composer)
    │   │   │       ├── node_modules/      # JS Dependencies (npm)
    │   │   │       ├── index.php          # Plugin Entry Point
    │   │   │       ├── composer.json
    │   │   │       ├── package.json
    │   │   │       ├── webpack.config.js
    │   │   │       ├── postcss.config.js
    │   │   │       └── README.md
    │   │   │
    │   │   ├── themes/
    │   │   └── uploads/
    │   │
    │   ├── wp-admin/
    │   ├── wp-includes/
    │   ├── wp-config.php
    │   └── index.php
    │
    ├── configs/                           # Server Configurations
    │   └── nginx/
    │
    ├── db/                                # MySQL Database Files
    │   ├── mysql/
    │   ├── performance_schema/
    │   ├── sys/
    │   └── wp/
    │
    └── docker-compose.yml                 # WordPress Docker Setup

Technology Stack

semantic-engine (Backend)

Runtime: Node.js
Web Server: Nginx (Reverse Proxy)
Vector Database: Qdrant
Container: Docker
AI/ML: Ollama (for embeddings and generation)

assistant-interface (Frontend)

CMS: WordPress
Backend Language: PHP
Frontend Framework: React (JSX)
Build Tool: Webpack
CSS: PostCSS
Dependency Management: Composer (PHP), npm (JavaScript)
Web Server: Nginx
Database: MySQL
Container: Docker

Communication Flow

User Input: User interacts with WordPress admin panel
Frontend Processing: React components handle user interactions
API Request: WordPress plugin sends HTTP requests to semantic-engine
Document Indexing: semantic-engine processes and stores documents in Qdrant
Query Processing: User queries are augmented and sent to Qdrant for semantic search
AI Generation: Retrieved context is used for RAG-based response generation
Response Delivery: Results are returned to WordPress and displayed to user

Development Setup

Prerequisites

Docker & Docker Compose
Node.js & npm
Composer
Ollama (for AI capabilities)

Getting Started

1. Start WordPress Environment

cd assistant-interface
docker-compose up -d --build

Access WordPress at: http://localhost:8000

2. Install Plugin Dependencies

# Inside the PHP container
cd wordpress/wp-content/plugins/assistant-interface
composer install
npm install
npm run build

3. Start RAG Backend

cd ai-assistant/semantic-engine
docker-compose up -d --build

4. Install Backend Dependencies

# Inside the engine container
npm install

API Endpoints

semantic-engine Endpoints

POST /index/store - Index new documents
DELETE /index/delete - Delete documents
POST /index/re-store - Re-index documents
POST /retrieval - Query documents (RAG)
GET /retrieval/document - Get specific document
GET /retrieval/documents - Get all documents
GET /retrieval/points - Get vector points

Key Features

Document Management

Upload and index documents
Delete and re-index capabilities
Vector-based storage in Qdrant

Semantic Search

Context-aware query processing
Prompt augmentation for better results
Vector similarity search

WordPress Integration

Native WordPress admin interface
React-based interactive UI
RESTful API integration

Name		Name	Last commit message	Last commit date
Latest commit History 21 Commits
assistant-interface		assistant-interface
semantic-engine		semantic-engine
.editorconfig		.editorconfig
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

AI Assistant

Architecture Overview

Conversation flow

Flow Diagram

Benefits of This Approach

Project Structure

Technology Stack

semantic-engine (Backend)

assistant-interface (Frontend)

Communication Flow

Development Setup

Prerequisites

Getting Started

1. Start WordPress Environment

2. Install Plugin Dependencies

3. Start RAG Backend

4. Install Backend Dependencies

API Endpoints

semantic-engine Endpoints

Key Features

Document Management

Semantic Search

WordPress Integration

About

Uh oh!

Languages

License

AbmSourav/ai-assistant

Folders and files

Latest commit

History

Repository files navigation

AI Assistant

Architecture Overview

Conversation flow

Flow Diagram

Benefits of This Approach

Project Structure

Technology Stack

semantic-engine (Backend)

assistant-interface (Frontend)

Communication Flow

Development Setup

Prerequisites

Getting Started

1. Start WordPress Environment

2. Install Plugin Dependencies

3. Start RAG Backend

4. Install Backend Dependencies

API Endpoints

semantic-engine Endpoints

Key Features

Document Management

Semantic Search

WordPress Integration

About

Topics

Resources

License

Uh oh!

Stars

Watchers

Forks

Languages