Prismatic Softwares Pvt. Ltd.

VAES – Voice-Assisted Expert System for IS Codes

Introduction

VAES is an intelligent voice-assisted chatbot designed to provide expert guidance based on Indian Standard (IS) codes. The system interprets user queries and retrieves relevant information from structured PDF documents containing IS codes.

Architecture Overview

Frontend

Built using Angular

User-friendly interface
Voice and text-based queries
Real-time conversational display

Backend

Developed in Python

Chat Interface APIs
RAG Pipeline
Embedding & Vector Store

LlamaIndex plays a crucial role in bridging structured PDFs and the retrieval engine, enabling efficient document processing and contextual response generation.

Core Components

Document Ingestion

Extracts and preprocesses content from PDF documents containing IS Codes

Embedding Generation

Converts text segments into dense vector representations for semantic search

Vector Storage

Manages high-dimensional embeddings with fast nearest-neighbor search

Interface Features

Voice Input

Natural voice-based interaction

Smart Responses

Context-aware conversational AI

Real-time Processing

Instant document retrieval

Learning System

Adaptive to user patterns

Conclusion

VAES proves that a domain-specific AI assistant can make complex technical standards accessible through natural conversation. By leveraging LlamaIndex, vector search, and generative AI, it bridges structured knowledge and user-friendly experiences.