# 👻 The Ghost That Lives Inside My Browser — AI-Powered Virtual Assistant

Sruti Gupta — Fri, 30 Oct 2026 06:37:24 +0000

This is my submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine.

What I Built

What if your browser had a personal assistant that could listen to you, understand your instructions, and actually help you get things done?

That's the idea behind my AI-Powered Virtual Assistant.

I built a web-based virtual assistant that brings voice interaction and AI-powered assistance together in one application.

Instead of typing every instruction, users can communicate with the assistant using their voice. It supports speech-to-text and text-to-speech, allowing users to interact naturally.

Some of its key features include:

🎙️ Voice-based interaction using the Web Speech API.
🤖 AI-powered responses using the Gemini API.
🔍 Search functionality.
▶️ YouTube search.
📍 Google Maps integration.
🌦️ Weather-related queries.
🧮 Mathematical calculations.
👤 Personalized assistant profiles with custom names and avatars.
🔐 Secure authentication using JWT.

I interpreted the theme Ghost in the Machine as giving a machine a voice and personality — something that feels less like a traditional website and more like an assistant you can actually talk to.

Demo

🚀 Live Demo: https://assitant-gamma.vercel.app/

Code

💻 GitHub Repository: https://github.com/SrutiGupta/virtualassistant

Feel free to explore the code and share your feedback!

How I Built It

I built this project using the MERN stack:

Frontend: React.js
Backend: Node.js and Express.js
Database: MongoDB
AI Integration: Gemini API
Voice Interaction: Web Speech API
Authentication: JWT
Media Management: Cloudinary

The Web Speech API handles speech recognition and speech synthesis, while Gemini provides the AI-powered conversational responses.

The application connects these components to create a smooth interaction between the user and the assistant.

A note on the AI model: The current version uses Google's Gemini API rather than a locally hosted open-weight model. The application is built with an open web-development stack, but I don't want to misrepresent the model itself as open-source.

Why Does Open Innovation Matter?

Open innovation makes it possible for developers to experiment, learn, build, and improve ideas without having to create every component from scratch.

For this project, open technologies and developer tools helped me bring together frontend development, backend services, authentication, voice interaction, and AI integration.

It also makes the project easier for other developers to explore, understand, extend, and contribute to.

I believe the best part of open innovation is that an idea doesn't have to remain limited to the person who first built it. Someone else can take it, improve it, and create something even better.

The Spirit of It 👻

The most interesting part of this project was seeing a simple voice command turn into an actual interaction.

I started with the idea of building an assistant that could respond to users, but along the way, I got to explore how voice technologies and AI can make web applications feel more interactive.

It reminded me that sometimes the most exciting part of development is making something that once felt futuristic work inside an ordinary browser.

And that's my little ghost in the machine! 👻

Prize Categories

AI-powered application

Thanks to the Hacktoberfest and DEV communities for creating opportunities like this to experiment, build, and learn.

Happy Hacking! 🚀

AI-Powered Virtual Assistant: Let AI Handle the Repetitive Stuff 🎙️🤖

Sruti Gupta — Sun, 25 Oct 2026 07:09:49 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

What I Built

I built an AI-Powered Virtual Assistant to make everyday digital tasks easier through voice commands.

We often repeat small tasks throughout the day, such as searching the web, opening YouTube, checking the weather, or performing calculations. Although these tasks are simple, switching between applications and typing commands repeatedly can take unnecessary effort.

My goal was to make these interactions faster and more convenient.

Here are some of the key features:

🎙️ Voice Interaction: Use voice commands instead of typing.
🧠 AI-Powered Intent Understanding: Uses the Gemini API to interpret requests and route commands.
🔎 Everyday Task Automation: Supports web search, YouTube, Maps, weather, and calculations.
🗣️ Speech Recognition and Synthesis: Enables voice input and spoken responses through browser speech technologies.
👤 Personalization: Customize the assistant's name and avatar.
🔐 Authentication and Profiles: Uses JWT authentication and MongoDB to support user accounts and persistent assistant profiles.

The idea is simple: let the assistant handle repetitive interactions so users can spend less time navigating applications and more time on what matters.

Demo

🔗 Live Demo: https://assitant-gamma.vercel.app/

Code

💻 GitHub Repository: https://github.com/SrutiGupta/virtualassistant

Feel free to explore the code and the implementation.

How I Built It

I built the application using the MERN stack, integrating AI capabilities and browser-based voice technologies.

MongoDB: Stores user and assistant profile data.
Express.js and Node.js: Handle backend APIs and application logic.
React: Powers the user interface.
Gemini API: Interprets user requests and generates structured command responses.
Web Speech API: Supports speech recognition and text-to-speech.
Cloudinary: Handles assistant avatar uploads.
JWT: Supports user authentication.

How It Works

The user speaks a command.
Speech recognition converts the voice input into text.
Gemini interprets the request and identifies the intended action.
The application routes the command to the appropriate functionality.
The assistant responds to the user.

This project combines conversational AI with practical full-stack development to create a more convenient way to interact with everyday digital tools.

Why Does Open Innovation Matter?

Open innovation gives developers more opportunities to experiment, customize AI systems, and choose technologies that fit their needs.

For a voice assistant like this, open-weight models could make it possible to explore local inference, reduce dependence on external AI services, improve control over data processing, and compare different models for specific tasks.

These possibilities are particularly interesting for assistants that handle personal workflows, where privacy, flexibility, and cost can matter.

My current implementation uses the Gemini API. Exploring an open-weight model in a future version would be a way to investigate these benefits further.

The Lazy Payoff 🦥

The biggest benefit is reducing the effort involved in everyday digital tasks.

Instead of manually typing a query or navigating between applications, users can speak a supported command and let the assistant handle the next step.

I haven't formally measured the time saved yet, but the goal is to make routine interactions more convenient, one voice command at a time.

Building this project reinforced an idea I really like: AI doesn't always need to solve a massive problem to be useful. Sometimes, making everyday tasks a little easier is a great place to start.

Why do everything manually when you can make technology do some of the work? 🚀

devchallenge #hf26challenge

DEV Community: Sruti Gupta

# 👻 The Ghost That Lives Inside My Browser — AI-Powered Virtual Assistant

What I Built

Demo

Code

How I Built It

Why Does Open Innovation Matter?

The Spirit of It 👻

Prize Categories

AI-Powered Virtual Assistant: Let AI Handle the Repetitive Stuff 🎙️🤖

What I Built

Demo

Code

How I Built It

How It Works

Why Does Open Innovation Matter?

The Lazy Payoff 🦥

devchallenge #hf26challenge