SwasthyaCall: Making Healthcare Accessible Through Voice AI

A developer built SwasthyaCall, an open-source voice-first healthcare appointment platform that lets patients book clinic appointments by phone call without an app, account, or smartphone. The system uses the open-weight Gemma 4 model to interpret natural-language requests, including Hindi, and extract structured booking data, paired with a full-stack web portal and a MongoDB data layer separating UI, AI interpretation, appointment logic, and storage. The developer notes AI-generated code still needs careful review for appointment conflicts, authentication, and healthcare safety flows.

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady https://dev.to/challenges/hacktoberfest-week3-2026-10-19 . What if someone could book a doctor's appointment without downloading an app, creating an account, or even having a smartphone? That question inspired me to build SwasthyaCall , an AI-powered healthcare appointment platform that makes accessing healthcare services as simple as speaking over a phone call. SwasthyaCall bridges the digital healthcare divide through voice-first AI. Many people, particularly elderly citizens and people in rural communities, face barriers when using digital healthcare applications. Internet access, language barriers, and unfamiliar interfaces can make booking an appointment unnecessarily difficult. SwasthyaCall provides a voice-based experience alongside a web portal for appointment management. The goal is simple: make appointment booking easier for patients and reduce repetitive administrative work for clinic staff. 🌐 Live Demo: https://hackday-dun.vercel.app/ https://hackday-dun.vercel.app/ Explore the application and try its available features. 💻 GitHub Repository: https://github.com/chirag-1264/hackday https://github.com/chirag-1264/hackday The repository contains the project's source code and setup instructions. Contributions, feedback, and suggestions are welcome I'd especially appreciate feedback on multilingual voice interfaces, accessibility, and making AI-powered healthcare tools more reliable. SwasthyaCall combines an open-weight AI model with a full-stack web application and a voice-based interaction layer. I used Gemma 4 as the language-understanding component of the project. It helps interpret natural-language appointment requests and extract structured information for the booking workflow. For example, a patient might say: Mujhe kal subah bukhar ke liye doctor ko dikhana hai. The system interprets the request and uses the extracted information to support the appointment-booking workflow. The model integration is designed to work with compatible inference endpoints, depending on the configured deployment. I used GitHub Copilot as a development assistant while building the application. It helped support the coding workflow, explore implementation approaches, and work through repetitive development tasks. AI-generated code still requires careful review and testing, particularly for appointment conflicts, authentication, and healthcare-related safety flows. MongoDB provides the document database layer for application data, including clinics, doctors, appointments, and related records. MongoDB Atlas offers a managed cloud database option for deploying and scaling the application's data layer. Database-level constraints and appropriate backend validation help maintain appointment consistency. The architecture separates the user interface, AI interpretation, appointment logic, and data storage so each component can be improved independently. Open innovation matters because useful AI shouldn't be limited to teams that can afford expensive proprietary infrastructure or closed APIs. For SwasthyaCall, an open-weight model such as Gemma provides opportunities to experiment with different deployment options, customize model integration, and adapt the application to different requirements. Open-source development tools and a flexible database architecture also make it easier for developers to inspect the code, contribute improvements, and experiment with new features. Most importantly, healthcare accessibility is a problem worth solving collaboratively. Developers can help improve language support, accessibility, reliability, and usability for communities that are often overlooked by digital-first services. Open innovation doesn't automatically guarantee privacy, safety, or lower costs, but it creates opportunities to inspect, adapt, and improve the technology. I used AI-assisted development as part of the building process. I haven't embedded a DevRelay agent session in this post. If I create a shareable session, I'll add it here so readers can explore the development process. The biggest potential payoff is reducing the effort required to book and manage appointments. Instead of navigating several screens, a patient can express an appointment request in everyday language. Clinic staff can manage bookings through a unified workflow rather than manually reconciling separate processes. Voice-based access can also offer an alternative for people who aren't comfortable using smartphone applications. I haven't measured time savings in a formal user study yet, so I won't claim a specific number of minutes saved. My next step is to test the workflow with users and measure booking completion time, failed requests, and administrative workload. I'm highlighting these partner technologies in my submission: I'll select the corresponding categories available in the official challenge submission form and verify their eligibility against the challenge rules. There are several areas I'd like to improve: Building SwasthyaCall has been an opportunity to explore how open AI models, web technologies, and voice interfaces can work together to address a real accessibility problem. If you're interested in AI, healthcare technology, open source, or voice interfaces, I'd love to hear your feedback What would you improve first: multilingual understanding, voice interaction, or appointment reliability?