-
-
"KaamSaarthi: Voice-Powered Solutions for Every Worker."
Inspiration : As a quality engineer at GMP Reels India Pvt Ltd, I’ve faced the frustration of suppliers delaying invoices or challans, making it hard to verify materials and complete receiving inspection reports on time. This personal challenge inspired KaamSaarthi. I also noticed India’s 63 million informal workers—drivers, vendors, and cleaners—struggle with similar issues due to literacy and language barriers, losing track of earnings or failing to issue invoices. I wanted to create a voice-first AI assistant that empowers them (and addresses my workplace pain) by enabling quick invoice generation and income tracking with commands like “Invoice banao” or “Aaj ₹500 kamaye” in Hindi and English, breaking barriers and ensuring timely documentation.
What it Does : KaamSaarthi is a voice-first AI assistant that simplifies financial management for informal workers. It allows users to: Log income/expenses with voice commands like “Aaj ₹500 kamaye.” Generate invoices with phrases like “Invoice banao ₹2000 ka painting kaam ke liye.” Switch seamlessly between Hindi and English to suit their preferences. Though the MVP focuses on income and invoices, the vision includes expenses and tips.
How I Built It: We started with Bolt.new, leveraging its AI-assisted environment to set up an Expo/React Native project. I created App.tsx as the entry point, using expo-speech for voice output. The interface features a simple layout with buttons for income logging, invoice generation, and language switching, styled with StyleSheet. Initial plans included ElevenLabs for better voice quality, but time constraints led us to stick with expo-speech. Testing was done in Bolt.new’s web preview, with manual triggers simulating voice commands.
Challenges I Encountered : Developing KaamSaarthi presented several significant challenges. As an electrical engineer with no prior experience in app development or coding, I embarked on this journey with a steep learning curve. Relying solely on guidance from prompts and discussions with Bolt.new’s AI assistant, as well as support from ChatGPT and Grok, I acquired the necessary skills to conceptualize and initiate the project. The time constraint was a critical factor, as I began the development process just two to three weeks prior to the hackathon deadline. Balancing this endeavor with my professional commitments as a working engineer further limited my available time. Despite these obstacles, I dedicated myself fully, working diligently until the final moments to achieve the maximum progress possible under the circumstances.
What's Next for KaamSaarthi: Next, I plan to resolve the current setup issues, fully integrate expo-speech for a smoother demo, and add expense tracking and daily tips. I’ll explore ElevenLabs or expo-speech-recognition for true voice input. Long-term, I aim to deploy KaamSaarthi with Supabase for data storage, expand to more languages, and adapt it for quality engineers like me to manage supplier invoices, empowering both informal workers and professionals with timely documentation.
Log in or sign up for Devpost to join the conversation.