🧠 Project Mani

The AI brain of RSMK Technologies — a smart, fast, and free personal AI assistant that knows everything about RSMK Technologies and powers intelligent chat across all RSMK websites.

---

What is Mani?

Project Mani is a centralized AI assistant built and owned by RSMK Technologies. It is not a generic chatbot — it is a purpose-built AI that carries deep knowledge about RSMK Technologies, its founder, and all projects under the brand.

Mani is designed to:

• Answer questions about RSMK Technologies and its projects accurately
• Power chat widgets embedded across multiple RSMK websites
• Use site-specific knowledge (RAG) to answer context-aware questions per website
• Fall back to general AI reasoning for queries outside its knowledge base

Think of Mani as the single AI brain that all RSMK websites share — each website adds its own layer of knowledge on top.

---

Architecture

┌─────────────────────────────────────────────────────┐
│                  RSMK Websites                      │
│   rsmk.me   │   BudgetBuddy   │   GridForge   │ ... │
└──────┬───────────────┬─────────────────┬────────────┘
       │               │                 │
       ▼               ▼                 ▼
┌─────────────────────────────────────────────────────┐
│              Mani Core API (Render)                  │
│                                                     │
│  ┌─────────────────────────────────────────────┐   │
│  │            Groq Inference Engine             │   │
│  │         llama-3.3-70b-versatile              │   │
│  └─────────────────────────────────────────────┘   │
│                                                     │
│  ┌──────────────┐    ┌──────────────────────────┐  │
│  │  RSMK Core   │    │   Per-Site RAG Context   │  │
│  │  Knowledge   │    │  (injected per request)  │  │
│  │    Base      │    │                          │  │
│  └──────────────┘    └──────────────────────────┘  │
└─────────────────────────────────────────────────────┘

How it works

1. A user asks a question on any RSMK website
2. The chat widget sends the query + site context to Mani Core API
3. Mani uses its RSMK knowledge base + the site's RAG context to answer
4. The response is returned and displayed in the chat widget

---

Project Structure

project-mani/
├── backend/                  → Mani Core API (Node.js + Express)
│   ├── src/
│   │   ├── groq.js           → Groq inference caller
│   │   ├── knowledge.js      → RSMK knowledge base loader
│   │   └── prompt.js         → System prompt builder
│   ├── knowledge/
│   │   ├── rsmk-core.json    → RSMK Technologies + founder info
│   │   ├── projects.json     → All RSMK projects data
│   │   └── achievements.json → Hackathons, certifications, etc.
│   ├── index.js              → Express server entry point
│   ├── .env.example
│   └── package.json
│
├── widget/                   → Embeddable chat widget
│   ├── mani-widget.js        → Drop-in script tag widget
│   └── mani-widget.css       → Widget styles
│
└── README.md

---

Tech Stack

Layer	Technology	Why
AI Model	Groq — llama-3.3-70b-versatile	Free, fastest inference, highly accurate
Backend	Node.js + Express	Lightweight, easy to maintain
Hosting	Render (free tier)	Always-on, no cold start issues
Knowledge	Structured JSON files	Simple, fast, no vector DB needed in v1
RAG	Context injection per request	Site-specific knowledge per website
Widget	Vanilla JS + CSS	Embeds into any website with one script tag

---

Phases

✅ Phase 1 — Mani Core

A standalone API that knows everything about RSMK Technologies.

• Groq-powered backend
• RSMK knowledge base (JSON)
• System prompt with Mani's identity
• Single /api/chat endpoint
• Deployed on Render

🔄 Phase 2 — Per-Site RAG Layer

Each website gets its own knowledge layer injected into Mani.

• rsmk.me → portfolio, skills, contact info
• BudgetBuddy → app features, FAQs, how-to guides
• GridForge → project docs, simulation info

🔜 Phase 3 — Embeddable Widget

A single script tag that any RSMK website can drop in to get a full Mani chat UI.

<script src="https://mani-core.onrender.com/widget/mani-widget.js"
        data-site="rsmk-portfolio">
</script>

🔜 Phase 4 — Ecosystem Integration

• Mani connects to BudgetBuddy data
• Mani connects to GridForge simulation outputs
• Project Mani dashboard for managing knowledge bases

---

API Reference

Base URL

https://mani-core.onrender.com

Health Check

GET /

{ "status": "Mani Core is live 🧠" }

Chat Endpoint

POST /api/chat

Request body:

{
  "query": "What is BudgetBuddy?",
  "siteContext": "User is on the BudgetBuddy app page.",
  "history": [
    { "role": "user", "content": "Hello" },
    { "role": "assistant", "content": "Hi! I'm Mani..." }
  ]
}

Response:

{
  "success": true,
  "response": "BudgetBuddy is an Android expense tracking app built by RSMK Technologies...",
  "model": "llama-3.3-70b-versatile"
}

---

Environment Variables

Variable	Description
GROQ_API_KEY	Groq API key from console.groq.com
PORT	Server port (default: 3001)

---

Local Development

# Clone the repo
git clone https://github.com/YOUR_USERNAME/project-mani.git
cd project-mani/backend

# Install dependencies
npm install

# Setup environment
cp .env.example .env
# Add your GROQ_API_KEY to .env

# Run the server
node index.js

# Test
curl -X POST http://localhost:3001/api/chat \
  -H "Content-Type: application/json" \
  -d '{"query": "Who is Mani?"}'

---

Deployment

Service	Platform	URL
Mani Core API	Render	https://mani-core.onrender.com
Widget CDN	Render static	https://mani-core.onrender.com/widget/

---

Calling the API (Frontend Integration)

Any authorized RSMK website or frontend (like mani-ui) can connect to Mani with a straightforward fetch call to the /api/chat endpoint.

Here is a quick example of how you send a question and get an answer:

const chatWithMani = async (userMessage, conversationHistory = []) => {
  try {
    const response = await fetch("http://localhost:3001/api/chat", { // Or your production URL
      method: "POST",
      headers: { "Content-Type": "application/json" },
      body: JSON.stringify({
        query: userMessage,
        // Optional context about the site the user is chatting from
        siteContext: "User is browsing the main RSMK ecosystem interface.",
        // History ensures Mani remembers the context of the conversation
        history: conversationHistory 
      })
    });

    const data = await response.json();
    
    if (data.success) {
      console.log("Mani says:", data.response);
      return data.response;
    } else {
      console.error("Mani Error:", data.error);
    }
  } catch (error) {
    console.error("Network Error:", error);
  }
};

Note on CORS: Ensure your frontend URL is allowlisted in the backend cors configuration, or the request will be blocked.

Detailed Backend Documentation

For more in-depth information about backend environment setup, API endpoint structure, and exact response formats, please refer to the Backend README.

---

Built by

RSMK Technologies — a personal tech brand by Srinivasa Manikanta Rajapantula.

rsmk.me · GitHub · LinkedIn