Where developers grow together

AutoResearch AI

AutoResearch AI is an autonomous multi-agent system that transforms fragmented startup inputs — scattered notes, PDFs, screenshots, and meeting transcripts — into a professional, investor-ready Business Requirements Document (BRD) in under 60 seconds. The system addresses Problem Statement #21: building a scalable, multi-modal AI system using Google Gemini AI and Google Cloud tools that processes real-time fragmented data and generates accurate, context-aware, explainable decisions. HOW IT WORKS:4 AI agents run autonomously in a LangGraph pipeline: 1. Research Agent — Extracts structured insights from all inputs2. Market Analysis Agent — Benchmarks competitors and sizes the market3. BRD Agent — Writes the full professional document with citations4. Architect Agent — Generates a Google Cloud system design blueprint Every claim in the output is source-cited, making the AI fully explainable and auditable. KEY FEATURES:- Multimodal input — PDFs, images, text, voice transcripts- Auto-generated SWOT analysis and competitor benchmarking- Feature matrix with P0/P1/P2 prioritization- Pricing strategy and go-to-market recommendations- BigQuery audit trail for enterprise-grade compliance- Real-time agent progress streaming via SSE- Export to PDF GOOGLE CLOUD INTEGRATION:- Gemini 1.5 Pro — Core reasoning and multimodal understanding- Vertex AI — Production model deployment and pipelines- BigQuery — Session audit logging and explainability metadata- Cloud Storage — Multimodal file ingestion IMPACT:Founders currently spend 40+ hours and ₹1,00,000 on consultants to produce a BRD. AutoResearch AI delivers the same output in 60 seconds, democratizing access to professional product strategy for early-stage startups — especially in Tier-2 and Tier-3 cities. LIVE DEMO:Frontend: autoresearch-emi1zjhkd-shresthbhargavas-projects.vercel.appBackend:  shresth0-autoresearch-ai.hf.space/docs TECH STACK:Google Gemini 1.5 Pro · Vertex AI · BigQuery · Cloud Storage · LangGraph · LangChain · FastAPI · Next.js 14 · TypeScript · Tailwind CSS · Docker · Hugging Face Spaces · Vercel Built by Team Vertex — Shresth Bhargava & Pranav Shyam NairHack Days Delhi · Major League Hacking · PS21

AsuraAi BRD

Nothing To Write Because my fellow teammate accedentaly delete it :( Ignore the description the real power dont comes from description , it comes from the LIve view   ignore this some random txt to fill desc===== blahblahblahblaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaablaaaaaaaaaaaaajj

Smart travelling companion

Project Title: Happy Routes AI (Premium Travel Intelligence & Adaptive Wearable Node) Happy Routes AI is an advanced, hardware-integrated IoT and Artificial Intelligence ecosystem designed to provide a highly personalized, secure, and automated travel experience for tech-savvy commuters and travelers. Developed as an end-to-end prototype, the system solves two major pain points in modern transit: complex itinerary planning and real-time solo travel safety. At its core, the platform utilizes a Flask-powered web backend coupled with a SQLite database architecture to manage user data, responsive wallet frameworks, and transaction ledgers. The software seamlessly orchestrates multi-modal travel workflows by processing environment snapshots and voice tokens through the Gemini AI Core Engine. Travelers can interact using natural language queries via a specialized Voice Assistant interface (leveraging browser-safe speech-to-text and text-to-speech pipelines) to dynamically map localized travel itineraries, book eco-optimized accommodations, or purchase transit corridors that automatically deduct from a synchronized secure wallet ledger. What elevates the project is its tight hardware-software synchronization. The platform interfaces with an external hardware wearable node via serial telemetry parsing protocols over microcontrollers (such as an Arduino or ESP32) equipped with an I2C OLED display. The wearable acts as a real-time monitor, flashing transaction acknowledgments (e.g., "Hotel Booked" strings) and continuously streaming kinetic vector coordinates back to the Flask application. Safety is managed through an automated Fall Detection and SOS Recovery Pipeline. Using real-time stream buffers, the backend continuously calculates vector magnitudes from the wearable’s accelerometer. If an impact anomaly passes a critical G-force threshold, a strict emergency modal with an active 10-second countdown is triggered on the dashboard. If un-dismissed, the system initiates an automated security dispatch, transmitting precise localized telemetry and alerts directly to predefined guardians via automated SMTP server routes. Happy Routes AI represents the future of contextualized, safe, and intelligent travel prototyping.