VineBot — Ministry AI Assistant

VineBot is an chatbot agent developed to serve the Blessed Vineyard Ministry. The goal was to create an accessible, digital spiritual companion that can answer questions regarding the ministry's vision, worship practices, and biblical teachings in real-time.

The core intelligence is powered by a Fine-Tuned Mistral 7B model integrated with Retrieval-Augmented Generation (RAG). This ensures that the bot doesn't just generate generic text, but provides accurate, context-aware answers based on specific ministry documents and scriptures.

The dataset used is the Ministry Manual of Blessed Vineyard Ministry.

---

🤖 Intelligent Chat Interface

The user interface is designed for focus and clarity, featuring a modern dark-themed design. Upon logging in, users are greeted with suggested prompts to help them get started, such as:

• "What is the vision of the ministry?"
• "What is the product ministry?"
• "How to do the activities and events?"

The system also features Chat History Management, allowing users to Rename or Delete previous conversations, ensuring a personalized and organized experience.

---

🧠 RAG-Powered Responses

import faiss
DATA_DIR = os.path.join(os.path.dirname(__file__), "data")
INDEX_PATH = os.path.join(DATA_DIR, "******.index")
DOCS_PATH  = os.path.join(DATA_DIR, "******.pkl")

DATA_DIR = os.path.join(os.path.dirname(__file__), "data")
INDEX_PATH = os.path.join(DATA_DIR, "******.index")
DOCS_PATH  = os.path.join(DATA_DIR, "******.pkl")

embedder = SentenceTransformer('all-MiniLM-L6-v2')  

index = faiss.read_index(INDEX_PATH)

with open(DOCS_PATH, "rb") as f:
    docs = pickle.load(f)


def retrieve_context(query: str, top_k: int = 3) -> str:
    q_vec = embedder.encode([query]).astype(np.float32)
    D, I = index.search(q_vec, top_k)
    chunks = [docs[i] for i in I[0] if 0 <= i < len(docs)]
    return "\n".join(chunks) if chunks else "No relevant context found."

bnb_config = BitsAndBytesConfig(
    load_in_8bit=True,
    llm_int8_enable_fp32_cpu_offload=True
)

Unlike standard chatbots, VineBot utilizes RAG (Retrieval-Augmented Generation). When a user asks a theological or organizational question, the system:

1. Retrieves relevant context from the ministry's knowledge base.
2. Generates a coherent, spiritually grounded response using the Mistral 7B model.

This ensures that answers—like the definition of the ministry's vision shown above—are consistent with the church's specific doctrines and values.

---

🔐 Secure Authentication

Security and identity management are handled via a robust authentication system. Users can access the platform using:

• Google OAuth Integration for one-click sign-in.
• Standard email/password credentials.
• A secure "Sign Up" flow for new community members.

The quoting of Ephesians 4:12 on the login screen sets the tone for the application, reminding users of the tool's purpose: "To equip his people for works of service."

---

👥 Admin & User Management

Administrators have full control over the platform through a dedicated Admin Dashboard. Key features include:

• User Overview: View all registered users, their roles (Admin/User), and account status.
• CRUD Operations: Admins can Add, Edit, or Delete users as needed.
• Role Assignment: Manage permissions to ensure only authorized personnel have admin access.

The interface allows for quick updates to user details, ensuring the community database remains accurate and up-to-date.

---

⭐ Feedback & Ratings System

To ensure the AI provides high-quality responses, a Feedback Loop is integrated directly into the chat. After receiving an answer, users can rate their experience (1-5 stars) and provide text feedback.

Admins can monitor these ratings via the dashboard to:

• Identify areas where the AI needs re-training.
• Review user satisfaction.
• Ensure the bot remains helpful and accurate.

---

🛠️ Tools & Technologies

• Mistral 7B — The core Large Language Model (LLM), fine-tuned for better instruction following.
• RAG (Retrieval-Augmented Generation) — For connecting the AI to specific ministry data.
• Python/Django — Backend server for handling requests and user management.
• Neon Serverless Postgres — Database.
• Tailwind CSS — For the responsive, dark-mode UI.

---

🌟 Key Impact

VineBot represents the intersection of Faith and Technology. By implementing advanced AI techniques like RAG, this project moves beyond simple automation to create a tool that genuinely understands and serves the specific needs of the ministry, making spiritual guidance accessible 24/7.