DEV Community: Gunjan Bhandari

Ghost in the Machine.

Gunjan Bhandari — Tue, 27 Oct 2026 10:34:11 +0000

👻 Ghost in the Machine. — The Haunted Story Generator

*This is a submission for the [Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine]

What I Built

Ghost in the Machine is a spooky interactive website that brings horror stories to life through the theme of haunted machines, mysterious messages, and forgotten memories.

Users can enter their own story ideas, choose a horror mood, and generate creepy stories. The project combines a dark, neon-inspired interface with an offline story generator and an optional integration with Ollama for locally running open-weight AI models.

I chose this idea because machines are a big part of our lives, but imagining what happens when a machine starts behaving in mysterious ways makes technology feel a little spooky!

Demo

👻 Try Ghost in the Machine.:

https://bgunjan884-ux.github.io/-Ghost-in-the-Machine/

Code

💻 GitHub Repository:

https://github.com/bgunjan884-ux/-Ghost-in-the-Machine

The project is built using three separate files:

index.html — Website structure
style.css — Dark theme, animations, and visual design
script.js — Story generation and interactive features

How I Built It

I built Ghost in the Machine using HTML, CSS, and JavaScript.

HTML provides the website structure, CSS creates the spooky atmosphere with purple neon effects and floating animations, and JavaScript handles user input, horror moods, story generation, and copying stories.

The project includes a rule-based offline story generator and an optional integration with Ollama to run an open-weight language model locally.

The local AI integration is designed to generate original stories from user prompts without relying on a hosted AI API. It requires Ollama and a compatible model to be installed and running.

Why Does Open Innovation Matter?

Open innovation gives developers the freedom to experiment, customize technology, and build projects without depending entirely on closed AI services.

With locally running open-weight models, developers can explore AI while keeping prompts on their own machines instead of sending them to an external AI provider. They can also experiment with different models and change how their applications behave.

Another important part of this project is its offline story generator, which provides a basic experience without an internet connection or an AI model.

For me, this project is an opportunity to explore how open technologies can make AI more accessible, customizable, and privacy-friendly.

My Agent Session

I used AI assistance to help plan the project structure, write the HTML, CSS, and JavaScript, and prepare the project documentation.

An agent session recording can be added here if available.

The Spirit of It

Imagine opening an old computer late at night and seeing a message appear on the screen:

"YOU LEFT ME HERE."

That small moment inspired Ghost in the Machine.. I wanted to turn a simple website into an interactive experience where technology, imagination, and horror meet.

Sometimes, the creepiest stories begin with a machine that should have been switched off.

Prize Categories

This project explores the open-source AI theme through its optional local-model integration.

Add specific partner prize categories here only if the project meets their eligibility requirements.

Made with curiosity, code, and a little bit of darkness. 👻

LazyFlow-AI

Gunjan Bhandari — Thu, 22 Oct 2026 18:04:34 +0000

🦥 LazyFlow AI — Let the Sloth Do the Work!

🌿 What I Built

Meet LazyFlow AI, a sloth-themed productivity assistant built around one simple idea: Why make everyday tasks harder than they need to be?

We all have little tasks that consume our time and energy—writing reminders, organizing our day, figuring out what to do next, or planning time to relax.

LazyFlow AI is designed to make those tasks easier.

With a calming interface and a simple text input, users can ask for help with everyday tasks, get to-do list suggestions, draft polite messages, and discover ideas for a relaxing evening.

The project is built using three separate files: HTML, CSS, and JavaScript. It includes a demo mode and an optional integration with Ollama for running an open-weight AI model locally.

The goal? Spend less time on busywork and more time enjoying life. 🦥

🚀 Demo

Try LazyFlow AI here:

🌐 Live Website: https://bgunjan884-ux.github.io/LazyFlow-AI/

Explore the interface, try the example prompts, and see how LazyFlow AI helps turn everyday requests into manageable next steps.

The website currently includes predefined demo responses. Real AI-generated responses require setting up Ollama and a compatible local model.

💻 Code

Check out the source code and explore the project on GitHub:

🔗 GitHub Repository: https://github.com/bgunjan884-ux/LazyFlow-AI

The project uses three core files:

index.html — website structure and content.
style.css — the calming, responsive design.
script.js — interactive features, demo responses, and optional local AI integration.

The goal was to keep the project simple, understandable, and easy to modify.

🛠️ How I Built It

I built LazyFlow AI using HTML, CSS, and JavaScript, focusing on a clean interface, simple interactions, and a relaxing user experience.

The main features include:

Task assistance and simple planning suggestions.
Message and reminder drafting.
Relaxation activity suggestions.
Copy and clear functionality.
Responsive layouts for different screen sizes.
A demo mode for testing without an AI installation.

For the local AI option, the project uses Ollama to connect the JavaScript frontend to a compatible open-weight model running on the user's computer.

For example, a user can download the Llama 3.2 model using:

ollama pull llama3.2

Once Ollama is running and local browser access is configured, the application can send a task to Ollama's local API and display the model's response.

This integration is optional. The current demo mode uses predefined templates rather than a live AI model.

🌍 Why Does Open Innovation Matter?

For me, open innovation is about giving people more control over the technology they use.

🔒 Privacy and Control

With a locally running model, requests can be processed on the user's own computer instead of being sent to a hosted AI API. This can be particularly useful for people who prefer to keep their personal tasks under their own control.

💸 More Accessible AI

Running a model locally avoids hosted AI API charges, although it still requires suitable hardware and electricity. This makes experimenting with AI possible without depending on a paid inference service.

🔄 Freedom to Experiment

Open-weight models give developers the opportunity to explore different models, change prompts, and customize an assistant's behavior without rebuilding the entire application.

🧩 Learning Through Open Technology

By using familiar web technologies and an optional local AI integration, LazyFlow AI provides a small, approachable example of how AI-powered applications can be built and customized.

A closed API can be convenient, but a local, open-weight approach offers a different advantage: greater control over model selection and where inference takes place.

🤖 My Agent Session

This project currently uses a straightforward frontend and an optional local model connection rather than a separate autonomous agent framework.

I have not included an agent-session recording. If I record a development session later, I can add it here.

🦥 The Lazy Payoff

The idea behind LazyFlow AI is not to make people more productive every second of the day. It is to reduce the effort needed to get started.

Instead of staring at a blank page, users can get a first draft. Instead of feeling overwhelmed by a long list, they can start with a few manageable steps. And instead of overplanning their free time, they can get a simple suggestion and relax.

The project is an early prototype, so I am not claiming measured time savings yet. My goal is to make everyday tasks feel a little lighter and give people more freedom to spend their time as they choose.

Sometimes, doing less really is the smarter approach. 🌿

🏆 Prize Categories

My project focuses on the Slow and Steady theme, exploring how open-weight AI can help reduce everyday busywork.

I have not selected any additional partner prize categories.

💚 Final Thoughts

Building LazyFlow AI has been an opportunity to explore how a simple web application can offer a useful starting point for local AI integration.

I want to keep experimenting, improving the experience, and exploring how open-source AI can help people spend less time on repetitive tasks.

Because technology should not always ask us to do more.

Sometimes, it should help us slow down.

Slow down. Stay curious. Let the sloth do the work. 🦥🌱

Built with HTML, CSS, JavaScript, and an optional Ollama integration for the Hacktoberfest Open-Source AI Challenge — Week 3: Slow and Steady.

Touch Grass - AI

Gunjan Bhandari — Sat, 10 Oct 2026 17:36:41 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

I Touch Grass built AI, an outdoor adventure companion designed to help people spend less time on screens and more time in nature.

Touch Grass AI helps users discover outdoor activities, take daily nature challenges, track their outdoor progress, and generate personalized outdoor adventure plans.

The project includes:

Outdoor activity suggestions
A 20-minute screen-free challenge timer
An outdoor activity progress tracker
An AI-powered adventure planner
A responsive interface built with HTML, CSS, and JavaScript

The goal is simple: use technology to encourage people to disconnect from their screens and reconnect with the real world.

Demo

I built and tested the project locally.

Screenshots and a video demonstration can be added here.

Code

The project uses three separate frontend files:

index.html — Website structure
style.css — Responsive styling and user interface
script.js — Timer, activity tracking, and AI integration

GitHub repository: [Add your GitHub repository link here]

How I Built It

I built GreenBuddy AI using HTML, CSS, and JavaScript.

For the AI component, I chose Ollama to run an open-weight language model locally on my computer.

The AI planner is designed to generate personalized outdoor activity plans based on the user's mood, available time, and preferred location.

The project is designed to use Gemma through Ollama. The AI functionality requires the local model and Ollama service to be installed and running.

The timer and progress tracker work directly in the browser using JavaScript.

Why Does Open Innovation Matter?

Open innovation makes GreenBuddy AI more accessible, flexible, and privacy-friendly.

By using an open-weight model through Ollama, I can experiment with local AI inference instead of depending on a paid, closed AI API.

This approach offers several advantages:

Privacy: Prompts can be processed locally instead of being sent to a third-party AI provider.
Flexibility: I can experiment with different compatible models.
Accessibility: Local inference can avoid per-request API charges.
Learning: Working with open AI tools helps me understand how AI applications work beyond simply calling a hosted API.

The project also demonstrates that AI can be used to encourage healthier offline habits instead of keeping people glued to their screens.

My goal is to make the screen the starting point of an outdoor adventure, not the destination.

My Agent Session

Optional: Add a DevRelay agent session if available.

Prize Categories

Hacktoberfest Open-Source AI Challenge — Week 1: Touch Grass.

Thanks for checking out GreenBuddy AI! 🌿