Back To Projects
CASE STUDY

Multilingual RAG Assistant for Government Services: MOSD Oman Case Study

Arabic-English RAG chatbot with Arabic Sign Language (ArSL) recognition for inclusive citizen access to government policies. Deployed on-prem with full data residency.

GovernmentIndustryPublic ServicesIndustryAccessibilityIndustryRAGCapabilityAI ChatbotCapabilityAccessibilityCapabilityOn-Prem DeploymentCapability
MOSD Oman logo for Ministry of Social Development, Oman -Bilingual Policy Assistant case study
MOSD Oman
Globcom logo for Ministry of Social Development, Oman -Bilingual Policy Assistant case study
Globcom
Executive Summarymultilingual RAG assistant for government services

The Ministry of Social Development in Oman needed a secure, inclusive way for citizens to access policy and service information without depending only on manual support. MythyaVerse built a bilingual Arabic-English assistant with a RAG pipeline for policies, schemes, FAQs, and service definitions. The system also includes Arabic Sign Language input, human escalation, service request tracking through MOSD APIs, staff dashboards, and on-prem deployment for data residency. The work strengthened citizen access while giving ministry teams a more scalable support layer for recurring questions and service guidance.

Business Impact

The assistant created a faster and more accessible citizen-service channel while preserving human escalation for complex cases. MOSD gained a scalable way to handle recurring policy and service questions under data residency constraints.

1 / 2

Dedicated Video Page

MOSD Oman bilingual policy assistant case study video

A video walkthrough of the Arabic-English policy assistant and accessibility workflow built for MOSD Oman.

Watch Video

Overview

The Ministry of Social Development needed an intelligent system that could help citizens quickly understand policies, access service information, and reduce dependency on manual support channels. The goal was a multilingual platform with strong accessibility support, including Arabic Sign Language (ArSL) input, and the ability to escalate to human assistance when required.

Problem

Citizens needed easier access to complex policy and service information, while MOSD needed to reduce repetitive manual support without reducing accessibility.

Constraints

  • Support Arabic and English policy access with grounded answers.
  • Include an accessible input path for hearing-impaired citizens through Arabic Sign Language.
  • Keep sensitive government data within on-prem infrastructure.
  • Connect conversational support with service request tracking and human escalation.

Solution Architecture

  • Policy documents, schemes, FAQs, and service definitions are cleaned and indexed for retrieval.
  • A multilingual RAG layer grounds answers in MOSD-approved source material.
  • Arabic Sign Language recognition expands access beyond text and voice-only interaction.
  • Escalation, video call support, service APIs, and staff dashboards connect the assistant to real operations.

The Challenge

MOSD needed a single conversational system that could address multiple goals at once. Citizens needed easier access to policy information, guidance on social services, and help understanding government processes. The Ministry also wanted to reduce the burden on human support teams without lowering service quality. Key challenges included:

  • Large volumes of recurring queries
  • Complex policy documents across multiple categories
  • Limited accessibility for hearing-impaired citizens
  • Need for bilingual support and accurate retrieval
  • Requirement for secure, on-premises deployment
  • A future-ready architecture that can scale with new services

Objectives

  • Provide bilingual conversational access in Arabic and English
  • Reduce manual support load for common queries
  • Improve accessibility, including Arabic Sign Language input
  • Deliver high accuracy through a reliable RAG pipeline
  • Enable real-time tracking of service requests through MOSD APIs
  • Maintain data residency and security with on-prem hosting
  • Offer human escalation with both text and video call options

Our Approach

MythyaVerse engineered a complete end-to-end system combining multilingual RAG, sign language recognition, real-time service integration, and a user-centric conversational interface. The process began with mapping MOSD's policies, schemes, FAQs, guidelines, and public service definitions. Documents were cleaned, indexed, and prepared for retrieval with emphasis on accuracy and consistency. All components were deployed fully on-premises to meet data residency and security requirements.

Solution Delivered

  • Multilingual RAG Chatbot -High accuracy answers for policies, schemes, public services, and citizen queries in both Arabic and English.
  • Arabic Sign Language Input -Real-time ArSL recognition enabling hearing-impaired citizens to interact with government services.
  • Accessibility Focused UX -Clear layouts, simplified flows, and inclusive interaction paths designed for broad citizen use.
  • Human Escalation Layer -Built-in escalation with text and video call support for cases requiring personal assistance.
  • Integration with MOSD Services -Direct API connections allow citizens to track ongoing service requests from within the chatbot.
  • Unified Dashboard -A console for MOSD staff to track escalations, monitor conversations, and review insights.
  • Secure On-Prem Deployment -All components hosted within MOSD infrastructure with strict data residency guarantees.

Technologies Used

Next.js technology logoNext.js
FastAPI technology logoFastAPI
Docker technology logoDocker
Kubernetes technology logoKubernetes
PostgreSQL technology logoPostgreSQL

Related MythyaVerse Services

Build something similar

Build a secure government assistant

Talk to MythyaVerse about a production-grade AI, automation, or XR system with the right constraints designed from day one.

Explore More Projects