Agentic & GenAI Systems
Production deployments of multi-agent AI systems, voice-first mental wellness platforms, and enterprise-grade GenAI solutions serving 15,000+ users with 99.9% uptime.
ManSetu - AI Mental Wellness Platform
India's first voice-first AI mental wellness companion serving 5,000+ active users and 10,000+ students
Problem & Context
India faces a severe mental health crisis with a massive treatment gap. Young Indians (18-35) struggle with exam stress, family pressure, workplace anxiety, and societal expectations, but lack accessible, culturally-aware mental health support. Traditional therapy is expensive, stigmatized, and often unavailable in regional languages.
Key Challenge: Build a voice-first AI platform that provides 24/7 mental wellness support in Hindi, English, and Hinglish while maintaining privacy and cultural sensitivity.
Solution & Technical Approach
Voice-First AI Architecture
- • Real-time emotion analysis from voice tone and pitch
- • Multilingual support (Hindi/English/Hinglish)
- • 60-second voice message processing
- • Text alternatives for accessibility
Wellness Features
- • CBT (Cognitive Behavioral Therapy) guidance
- • Smart AI journaling with pattern recognition
- • Mood tracking and analysis
- • Breathing exercises and meditation
Privacy & Compliance
- • End-to-end encryption
- • Data hosted on Indian servers
- • DPDP Act 2023 compliant
- • Direct Tele-MANAS (14416) integration for crisis
Accessibility
- • Completely free for students
- • No credit card required
- • 24/7 availability across Indian time zones
- • Web and mobile applications
Technology Stack
Outcomes & Business Impact
Agentify - Multi-Agent AI Platform
Production-grade platform orchestrating 15+ specialized AI agents for enterprise application development
Problem & Context
Attri.ai needed a production-grade platform to orchestrate 15+ specialized AI agents for enterprise application development. The platform had to support complete SDLC automation from PRD generation to deployment, handle secure code execution in isolated environments, enable real-time multi-agent collaboration, implement multi-tenant SaaS with workspace-level billing, and provide comprehensive observability - all while maintaining 99.9% uptime with zero-downtime deployments.
Key Challenge: Build an enterprise-ready multi-agent system that automates the entire software development lifecycle while maintaining security, scalability, and reliability.
Solution & Architecture
15+ Specialized Agents
- • Orchestrator Agent: Workflow coordination
- • PRD Agent: Requirements generation
- • Solution Architect: System design
- • UI Designer: Interface creation
- • Coder Agent: Code generation
- • Diff Agent: Change management
- • Discovery Agent: Context analysis
Enterprise Features
- • Complete SaaS billing (Stripe integration)
- • Multi-tenant workspace management
- • Zero-downtime deployments
- • Real-time collaboration (Azure WebPubSub)
- • Comprehensive observability (Datadog)
- • Automated Vercel deployments
Security & Isolation
- • E2B MicroVMs for code execution
- • Sandboxed environments per agent
- • Secure API key management
- • Rate limiting and abuse prevention
Advanced Capabilities
- • MCP (Model Context Protocol) integration
- • RAG pipeline with pgvector
- • React Flow workflow visualization
- • Template generation system
Technology Stack
Outcomes & Business Impact
Explore More Projects
View additional AI/ML projects, open source contributions, and technical experiments