Astrix AI

Language is no longer Barriers

Comment

Inspiration

Across India's 29 states, millions of citizens face severe structural barriers when trying to access basic public services, legal assistance, and emergency care. The current public administration ecosystem is heavily burdened by linguistic exclusion—where regional non-English/non-Hindi speaking populations struggle to navigate digital platforms—as well as slow manual approval chains, paper dependency, and localized safety disparities between isolated villages and state capitals.

Astrix AI was inspired by the vision of a truly unified, paperless, and inclusive Digital India. We set out to bridge the urban-rural divide by combining localized cultural context, real-time emergency safety infrastructure, and proactive agentic AI that actively performs bureaucratic tasks on behalf of citizens.

What it does

Astrix AI is an AI-powered public digital infrastructure and emergency safety platform designed to streamline civic workflows and democratize public access:

  • Culturally Immersive Multilingual Access: Integrates the Bhashini API to power localized dashboards across regional languages (such as Punjabi and Tamil). It features interactive state maps and culturally grounded animations (e.g., Bhangra for Punjab) to make public technology intuitive and welcoming for rural citizens.

  • Autonomous Agentic Automation: Replaces passive forms with AI agents (powered by CrewAI/LangChain) that gather details via text or voice, auto-fill complex legal documents, and programmatically follow up on administrative approvals.

  • Paperless Governance Workflow: Fully digitizes paperwork processes to lower human error, eliminate bureaucratic delays, and reduce physical paper dependency for state and local departments.

  • Hyper-Local Geospatial Intelligence: Provides dynamic mapping that overlays precise court locations, police station jurisdictions, and administrative boundaries on a clear interactive dashboard.

  • Unified Low-Bandwidth Emergency SOS Network: Connects remote villages to state capitals via lightweight telemetry transmission, delivering instantaneous, reliable real-time SOS safety alerts even under poor cellular connectivity.

How we built it

We built Astrix AI using a full-stack, distributed microservices architecture tailored for speed, spatial awareness, and AI orchestration:

  • Frontend: Developed with Next.js (React) and Flutter, formatted in a high-contrast "Golden Bird" (golden-black) patriotic aesthetic. Styled with Tailwind CSS, with interactive animations driven by Lottie/Framer Motion and map renders managed by Leaflet.js / Mapbox GL JS.

  • Backend & AI Orchestration: Powered by FastAPI (Python) and Express.js (Node.js). We integrated the government-backed Bhashini API for real-time translation and speech processing, and orchestrated autonomous AI task execution using CrewAI and LangChain paired with models like Sarvam AI's Samudra and open-source LLMs.

  • Database Layer: Utilizes PostgreSQL with PostGIS for spatial and jurisdictional query boundaries, MongoDB / Redis for fast real-time session state management and SOS logging, and Pinecone / Qdrant as a high-performance vector store for legal document retrieval.

  • DevOps & Real-time Services: Hosted on AWS / Google Cloud Platform in Indian data centers. Real-time SOS alerts and continuous data feeds use WebSockets and Firebase Cloud Messaging inside Docker and Kubernetes containers.

Challenges we ran into

  1. Multilingual Context and Dialect Nuances: Integrating the Bhashini API across diverse Indian languages required extensive handling of localized terminology, especially when translating complex legal jargon into understandable regional phrasing.

  2. Autonomous Agent Reliability: Designing multi-agent flows via CrewAI/LangChain that could accurately process multi-step bureaucratic procedures without dropping context or producing hallucinated inputs in official forms.

  3. Low-Bandwidth Geospatial Telemetry: Ensuring the unified SOS safety network could instantly transmit accurate geo-coordinates from remote tier-3 villages over weak 2G/3G connectivity networks.

  4. Complex Spatial Indexing: Mapping irregular jurisdictional boundaries across multiple state departments into PostGIS while maintaining real-time map rendering performance on client devices.

Accomplishments that we're proud of

  • Proactive "Agentic" Execution: Moving beyond passive information chatbots to build autonomous AI agents that actually execute multi-step paperwork and track administrative pipelines.

  • Inclusive Cultural Interface: Designing an approachable interface that combines state-level cultural animations with Bhashini translation to eliminate technology intimidation for non-English speakers.

  • Asymmetrical Safety Network: Delivering the exact same high-priority, real-time emergency SOS infrastructure to isolated rural villages as to major metropolitan hubs.

  • Cohesive Tech Integration: Successfully stitching together vector databases, spatial mapping, translation APIs, and multi-agent frameworks into a unified, lightweight system.

What we learned

  • Design for Accessibility First: Rural digital adoption depends heavily on localized visual cues, voice interfaces, and low-friction interactions rather than text-heavy corporate UI layouts.

  • Agentic Workflows Require Strict Guardrails: Autonomous agents executing real-world administrative tasks need robust verification gates and fallback mechanisms to ensure 100% compliance with legal frameworks.

  • Public Infrastructure Needs Lightweight Data Pipelines: Building systems for nationwide reach requires optimizing payloads (like SOS telemetry) so that core safety services run seamlessly regardless of bandwidth constraints.

What's next for Astrix AI

  • Deep India Stack Integration: Direct integration with national digital utilities like DigiLocker and Umang for instant, verified identity checks and document pulls.

  • Edge-AI Offline Safety Protocol: Deploying quantized, lightweight local models to run directly on mobile devices, allowing emergency SOS signaling and basic guidance without active internet connections.

  • Municipal Micro-Bureaucracy Swarms: Expanding AI agent swarms to automate hyper-local civic services like municipal permits, public utility registration, and local grievance redressal.

  • Global Digital Public Infrastructure Expansion: Adapting the core Astrix AI framework into an exportable, open public digital infrastructure model for other developing nations seeking unified, paperless governance.

Built With

Share this project:

Updates

Submission history