DEV Community: Priyanshu saroj

PhantomAI – Intelligence Beyond the Unknown

Priyanshu saroj — Sat, 31 Oct 2026 14:27:53 +0000

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

👻 PhantomAI – Intelligence Beyond the Unknown

What I Built

I built PhantomAI, an AI-powered supernatural web experience inspired by the theme "Ghost in the Machine."

What if a mysterious intelligence lived inside your computer and could communicate with you?

PhantomAI brings this idea to life through an interactive AI ghost chatbot and a horror story generator.

Users can:

👻 Chat with a mysterious AI entity.
📖 Generate unique horror stories using custom prompts.
🌑 Explore a futuristic, dark-themed supernatural interface.
⚡ Experience AI powered by an open-weight language model.

My inspiration was to combine the mystery of Halloween with the possibilities of open-source artificial intelligence.

Demo

Project Name: PhantomAI

Website: PhantomAI – Live Demo

GitHub Repository: View PhantomAI Source Code

Code

Check out the complete source code:

🔗 GitHub – PhantomAI Repository

The project is built using:

HTML5 – Website structure.
CSS3 – Dark futuristic interface, glowing effects, and animations.
JavaScript – Interactive functionality and AI integration.
Ollama – Local AI model execution.
Qwen 2.5 – Open-weight language model.

How I Built It

I built PhantomAI using HTML, CSS, and JavaScript for the frontend, with Ollama providing the local AI inference.

For the AI component, I chose Qwen 2.5 (3B), an open-weight language model that runs locally through Ollama.

The application communicates with the Ollama API to generate responses.

The main AI features are:

1. AI Ghost Chatbot

Users can communicate with a fictional supernatural AI entity. The chatbot uses the local language model to generate dynamic responses.

2. Ghost Story Generator

Users can enter their own ideas, select a horror style, and generate original stories using AI.

3. Local AI Integration

Instead of depending on a paid, hosted AI API, PhantomAI sends prompts to a locally running Ollama model.

This makes the project more customizable and allows local inference without an internet connection after the model has been downloaded.

Why Does Open Innovation Matter?

Open innovation is at the heart of PhantomAI.

I wanted to explore how AI-powered applications could work without depending entirely on closed, proprietary AI services.

Using an open-weight model through Ollama provides several advantages:

Privacy: Prompts and conversations can remain on the user's local machine.
Accessibility: No paid hosted AI API is required.
Customization: Developers can experiment with different models and prompts.
Offline Potential: The AI can run without internet access after downloading the model.
Independence: Developers have more freedom to modify and experiment with their AI setup.

For me, open-source AI is not just about building something interesting. It is about having the freedom to understand, experiment, and create without being locked into a single provider.

The Spirit of It

Halloween is a time for mysteries, supernatural stories, and things we cannot explain.

With PhantomAI, I wanted to give that feeling a technological twist.

Instead of imagining a ghost in an abandoned house, I imagined one living inside a computer.

A machine that responds when you speak, creates stories from your imagination, and makes you wonder:

Is it just artificial intelligence, or is something else answering back?

That idea became PhantomAI.

👨‍💻 Built By

Priyanshu Saroj

GitHub: @priyanshusaroj385-boop

👻 PHANTOMAI
Intelligence Beyond the Unknown.

The machine never sleeps. Something is always listening.

PhantomAI 👻 | A Ghost in the Machine — Open-Source AI Experience

Priyanshu saroj — Tue, 27 Oct 2026 09:05:39 +0000

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

What I Built

👻 PhantomAI — Some Machines Remember What We Forget

PhantomAI is a spooky, futuristic website inspired by the Ghost in the Machine theme. It imagines a world where forgotten computers whisper mysterious messages, abandoned code holds untold stories, and every interaction feels like opening a door into the unknown.

I built PhantomAI using HTML, CSS, and JavaScript, combining a dark neon interface with two interactive experiences:

Ghost Story Generator: Enter a story idea and receive a spooky story.
Haunted Terminal: Send messages to a fictional haunted computer and receive mysterious replies.

The experience features a floating ghost, glowing purple and cyan accents, responsive layouts, and a terminal-inspired interface.

The current browser demo uses predefined stories and responses. I also explored an optional integration with Ollama to connect the interface to a locally running open-weight model.

The idea behind PhantomAI is simple: what if forgotten technology could tell us a story?

Demo

GitHub Repository:

https://github.com/priyanshusaroj385-boop/PhantomAI

Live Demo:

Not deployed yet. The project currently runs locally in a browser. A deployed demo can be added here once available.

Code

Check out the source code and project files:

🔗 PhantomAI on GitHub

The project uses:

index.html — Website structure and content.
style.css — Visual design, responsive layout, and ghost animations.
script.js — Story generation, terminal interactions, and demo responses.

The frontend is intentionally lightweight, with no frontend framework or paid API required for the demo experience.

How I Built It

I started with a simple idea: make an ordinary browser interface feel like a haunted machine.

1. Designing the experience

I created a dark interface using neon-purple gradients, cyan highlights, glowing borders, and a floating ghost illustration. The layout adapts to desktop and mobile screens.

2. Building the interactions

Using vanilla JavaScript, I implemented a story generator and a haunted terminal. The current version uses predefined content to demonstrate these interactions without requiring an API key.

3. Exploring local AI

I designed an optional integration with Ollama's local chat API. With Ollama running and a compatible model installed, the interface can be extended to generate responses locally.

This integration requires a working local service and appropriate browser access. It is not necessary for the basic demo.

Technology stack:

HTML5
CSS3
Vanilla JavaScript
Ollama for optional local inference
Llama 3.2 as an example model to explore

An important distinction: the submitted browser demo currently uses predefined responses. The optional Ollama integration is a separate capability that requires local setup; the project should not be described as already running AI inference unless that integration has been tested successfully.

Why Does Open Innovation Matter?

Open innovation gives developers the freedom to experiment instead of building everything around a single closed service.

For a project like PhantomAI, that freedom matters in three ways.

Privacy and control

With a properly configured local model, prompts can remain on the user's own computer rather than being sent to an external AI provider. This can be especially useful for creative writing and personal stories.

Freedom to experiment

Open-weight models give developers options to explore different models, adjust prompts, and customize behavior. The project can evolve without being permanently tied to one hosted AI API.

Accessibility

A browser-based interface can be built using free development tools, and local inference can avoid per-request hosted API charges. Running models still requires suitable hardware and electricity, and model licenses vary.

PhantomAI is an early experiment in this direction. Its current demo establishes the interface and interactions, while the local-model integration provides a path toward more private and customizable AI experiences.

For me, open innovation is about having the freedom to understand, change, and build on the technology instead of treating it as a black box.

My Agent Session

I don't have a DevRelay agent-session link to share for this submission.

This section can be updated if a session is recorded and made available.

The Spirit of It

Imagine opening an abandoned computer late at night.

The screen flickers. A terminal appears. You type a question, and a message returns from a machine that seems to remember something you don't.

That is the atmosphere I wanted PhantomAI to capture.

Beyond Halloween aesthetics, the project is also about preserving forgotten ideas. Open-source software can outlive its original creators, and accessible tools let new developers explore, adapt, and continue work that might otherwise be abandoned.

Sometimes, bringing something back to life simply means giving someone the freedom to build on it.

Prize Categories

Overall Hacktoberfest Open-Source AI Challenge entry.

No partner-specific prize category is claimed at this stage. I will add a partner category only if the project uses the corresponding technology in a qualifying way.

Built with curiosity, code, and a little mystery. 👻

PhantomAI — Open models. Forgotten stories. Unlimited imagination.

🦥 LazyBuddy AI: Work Less, Live More — Building for the Slow and Steady Open-Source AI Challenge

Priyanshu saroj — Sun, 25 Oct 2026 12:03:44 +0000

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

What I Built

🦥 LazyBuddy AI — Work Less, Live More

LazyBuddy AI is a simple productivity website designed to make everyday tasks easier. It helps users turn long text into shorter notes, create to-do checklists, and organize messy text.

The goal is simple: reduce repetitive work and give people more time to focus on what matters.

Features

📝 Quick text summarization
✅ Automatic checklist generation
📂 Text organization and duplicate removal
📋 Copy results with one click
📱 Responsive design for desktop and mobile
🔒 Browser-based text processing without sending text to a server

Demo

Run the project locally by opening index.html in your browser.

Local demo: Available on my computer.

A public live demo will be added when deployed.

Code

GitHub Repository: LazyBuddy AI

The project uses HTML, CSS, and vanilla JavaScript, with separate files for the website structure, styling, and functionality.

How I Built It

I built LazyBuddy AI using:

HTML5 for the structure
CSS3 for the design, responsive layout, and animations
JavaScript for text processing, checklist generation, and organizing notes

The current version uses browser-based JavaScript rules rather than an AI model. My next step is to integrate an open-weight model running locally through Ollama so that users can get more intelligent summaries and task suggestions without relying on a closed AI API.

Why Does Open Innovation Matter?

Open innovation gives developers the freedom to experiment, customize models, and build tools without depending entirely on proprietary AI services.

For LazyBuddy AI, a local open-weight model could help keep personal notes on the user's device, reduce dependence on paid APIs, and allow the application to work without sending prompts to an external AI provider.

It also gives developers more control over model selection, privacy, and future improvements.

I want to explore how open AI can make everyday productivity tools more accessible and customizable.

The Lazy Payoff

LazyBuddy AI aims to reduce the effort required to summarize notes, prepare task lists, and organize information.

Instead of manually cleaning up text or creating a checklist from scratch, users can paste their content and get a more organized starting point in seconds.

The current prototype demonstrates these basic workflows. I plan to measure actual time savings and improve the results as the project develops.

Prize Categories

This project currently demonstrates a browser-based productivity prototype.

I plan to explore the open-weight AI and local-inference aspects of the challenge by integrating a locally running model. I will select applicable partner prize categories after confirming that the implemented features meet their requirements.

HustleAI: An Open-Source AI Marketing Copilot for Small Sellers 🚀

Priyanshu saroj — Thu, 15 Oct 2026 09:54:00 +0000

HustleAI — Your AI-Powered Side Hustle Assistant 💻

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

I built HustleAI — an AI Marketing Copilot for Small Sellers, designed to help freelancers, independent creators, and small businesses turn product ideas into sales-ready content.

Creating marketing content consistently can be time-consuming, especially for people who don't have a dedicated marketing team or budget for expensive tools. HustleAI helps generate product descriptions, social media captions, and headlines based on product details, platform, tone, and content type.

The goal is to make AI-powered marketing more accessible, affordable, and privacy-conscious by supporting local open-weight AI models.

Key features:

Generate product descriptions, social media captions, and marketing headlines.
Customize content for different platforms and tones.
Use a simple, futuristic dashboard.
Connect to a locally running AI model rather than depending on a paid hosted AI API.
Keep the project open for experimentation and future improvements.

Demo

🚀 Live Website: https://priyanshusaroj385-boop.github.io/HustleAI/

Code

💻 GitHub Repository: https://github.com/priyanshusaroj385-boop/HustleAI.git

Explore the source code, run the project locally, and contribute improvements.

How I Built It

I built HustleAI using a web development stack focused on simplicity and accessibility:

HTML — structures the application.
CSS — creates the responsive, futuristic interface.
JavaScript — handles form interactions, content generation requests, and UI updates.
Node.js — provides the backend API.
Ollama — runs an open-weight AI model locally.
Qwen 2.5 3B — the intended default model for local text generation.

The intended AI workflow sends the user's product information to the local backend, which communicates with Ollama and returns generated marketing content. This avoids requiring a paid hosted AI API for inference.

The frontend is deployed on GitHub Pages, while local AI generation requires running the backend and Ollama in an appropriate environment. The public frontend alone does not provide a hosted AI backend.

Why Does Open Innovation Matter?

Open innovation makes AI experimentation more accessible to developers, students, and small businesses that may not have the budget for expensive APIs.

For HustleAI, open-weight models and local inference make it possible to:

Reduce recurring costs: Run supported models locally without paying a per-request inference fee.
Improve privacy: Keep prompts and generated content on the user's device when the application is configured for local inference.
Choose and replace models: Experiment with different open-weight models as requirements change.
Learn by building: Understand the complete workflow from a web interface to a backend and a locally running AI model.
Build independently: Develop and customize an AI-powered product without depending entirely on a closed AI service.

Open innovation is not just about using a free model; it is about giving developers more control over how AI products are built, tested, and improved.

My Agent Session

Optional — add your DevRelay agent session here if you have one.

Prize Categories

Add the exact partner prize categories you qualify for from the official challenge page.

Project: HustleAI

Theme: Side Hustle

Focus: Open-source AI, local inference, privacy-conscious marketing assistance, and accessible web development.

TouchGrass AI: An Open-Source AI-Powered Platform for Outdoor Adventures and Sustainable Digital Well-Being

Priyanshu saroj — Fri, 09 Oct 2026 11:20:00 +0000

🌿 TouchGrass AI

Your Offline Outdoor Adventure Companion

Live Website: https://priyanshusaroj385-boop.github.io/Touch-grass-ai/
GitHub Repository: https://github.com/priyanshusaroj385-boop/Touch-grass-ai

1. Project Overview

TouchGrass AI is a web-based outdoor adventure companion designed to encourage people to spend more time outdoors and reduce unnecessary screen time. It helps users discover outdoor activities, explore nature, maintain a nature journal, and get gardening ideas.

2. Problem Statement

Excessive screen time and limited outdoor engagement can reduce opportunities for physical activity, relaxation, and connection with nature. People may also struggle to find simple, accessible outdoor activities that suit their interests and available time.

3. Proposed Solution

TouchGrass AI provides a simple, interactive website that suggests outdoor activities based on users’ interests, mood, available time, and location. It also includes nature journaling, gardening suggestions, nature exploration, and outdoor checklists.

4. Key Features

Adventure Generator: Discover outdoor activities based on your preferences.
Nature Journal: Record observations and experiences from outdoor adventures.
Garden Assistant: Explore gardening ideas and plant-care tips.
Nature Explorer: Discover ways to observe and learn about nature.
Outdoor Checklist: Prepare for outdoor activities.
Responsive Design: Access the website on desktop and mobile devices.

5. Technology Stack

Frontend: HTML5, CSS3
Programming: JavaScript
Data Storage: Browser localStorage
AI Roadmap: Integration with open-source or open-weight AI models.

Note: The current prototype uses predefined JavaScript suggestions. Live AI integration is a planned enhancement.

6. How to Use

Open the live website.
Select your mood, available time, and preferred activity.
Generate an outdoor adventure suggestion.
Explore the nature journal, garden assistant, and checklist features.
Head outdoors and enjoy your activity!

7. Project Links

Live Demo: https://priyanshusaroj385-boop.github.io/Touch-grass-ai/

Source Code: https://github.com/priyanshusaroj385-boop/Touch-grass-ai

8. Future Scope

Integrate an open-source AI model for personalized recommendations.
Add plant identification and nature-learning features.
Improve offline functionality.
Add weather-aware activity suggestions.
Introduce progress tracking for outdoor habits.

9. Expected Impact

TouchGrass AI aims to encourage healthier digital habits, support outdoor exploration, and make nature-based activities easier to discover.

10. Author

Priyanshu Saroj
Project: TouchGrass AI