DEV Community: Swarit Mathur

SlothDesk AI — Your First File Organization Assistant

Swarit Mathur — Mon, 26 Oct 2026 06:58:47 +0000

What I Built
Meet SlothDesk AI 🦥 — a local-first file organization assistant that helps you spend less time sorting files and more time doing what actually matters.

Downloads folders have a habit of becoming digital junk drawers. Documents, images, archives, and miscellaneous files pile up until finding anything becomes a task of its own.

SlothDesk AI helps bring order to that chaos.

It scans a folder, analyzes file metadata, and builds an organization plan that you can review before applying any changes. It can also use optional, locally running Ollama models to suggest file categories and more meaningful filenames.

The principle behind it is simple: AI should do the tedious thinking, but you should stay in control.

What makes it different?

🌿 Local-first: Work with files on your own device.

🧠 Optional local AI: Get organization suggestions through Ollama without depending on a paid AI API.

🛡️ Human approval: Review the proposed changes before applying them.

📁 Safer organization: Designed to avoid accidental overwrites and file deletion.

⚡ Simple workflow: Scan, review, and organize through a Streamlit interface.

It's for anyone whose Downloads folder needs a little less chaos — students, developers, researchers, and anyone who would rather not sort files manually.

Demo
🎥 Demo video: Watch my SlothDesk AI walkthrough
https://drive.google.com/file/d/1WQiO4eIE6i3EJqLnfjQQDny6ANSVsF87/view?usp=drivesdk
The walkthrough demonstrates the app's interface and file-organization workflow.

Code
💻 GitHub repository:
https://github.com/SM01010011/SlothDesk-AI

The project is built as an open-source Python application. Explore the repository for the implementation, setup instructions, and tests.

How I Built It
SlothDesk AI combines a lightweight Python application with an optional local AI suggestion engine.

Tech stack:

Python — core application and file-organization logic

Streamlit — interactive user interface

Pandas — tabular presentation of scanned file information

Ollama — optional local inference for AI-powered organization suggestions

Pytest — automated testing

The workflow separates planning from execution. The application scans the selected folder and builds an organization plan first. The user can review that plan before choosing to apply it.

The AI component is optional. When enabled, it helps suggest categories and filenames using file metadata rather than sending file contents to the suggestion engine. This keeps the project focused on practical assistance without giving an AI model unrestricted control over the filesystem.

I wanted the experience to feel useful even without a cloud AI subscription. Local inference makes that possible for users with a compatible Ollama setup and model.

Why Does Open Innovation Matter?
For a project like SlothDesk AI, open innovation isn't just about making the source code visible. It changes who can use, inspect, adapt, and improve the tool.

Using open-source tools and local inference makes experimentation possible without requiring a paid proprietary AI API. Developers can inspect the implementation, run it in their own environment, test different compatible models, and adapt the workflow to their needs.

It also makes privacy-conscious experimentation more accessible. Instead of building the entire experience around sending data to a remote service, SlothDesk AI can use local models for its optional suggestions.

There is still work to do: local inference depends on the user's hardware and installed model, and AI-generated suggestions should always be reviewed. But open tools make it possible to build and improve this kind of practical assistant with fewer external dependencies.

That's the kind of open innovation I wanted to explore: use AI to reduce repetitive work without sacrificing user control.

The Lazy Payoff
The goal is to reduce the repetitive effort involved in sorting a cluttered folder.

Rather than manually deciding where each file belongs and thinking up better names one at a time, users can scan a folder, review suggested organization, and approve a plan.

I haven't measured a verified time-saving figure yet, so I won't claim a specific number of minutes saved. The next step is to test it on real, cluttered folders and measure how much manual effort the workflow actually removes.

Slow and steady — build something useful, test it honestly, and keep improving it. 🌱

Prize Categories
This submission is entered for the overall Hacktoberfest Open-Source AI Challenge: Week 3 — Slow and Steady.

SlothDesk AI uses optional local inference through Ollama to support privacy-conscious file organization without requiring a paid proprietary AI API.

PhishBrief AI: Private, Local-First Phishing Analysis with Open-Weight AI

Swarit Mathur — Tue, 13 Oct 2026 15:04:20 +0000

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

What I Built
PhishBrief AI is a local-first AI-powered phishing analysis tool designed to help small businesses and individuals understand suspicious emails and messages before interacting with them.

Phishing emails often use urgency, impersonation, suspicious links, and requests for sensitive information to trick people. Many small teams lack dedicated security staff or enterprise-grade email security tools.

PhishBrief AI aims to make initial phishing analysis more accessible by helping users examine suspicious content, understand potential warning signs, and receive actionable security recommendations.

The project explores a privacy-conscious approach: running an open-weight AI model locally instead of requiring every message to be sent to a hosted AI API.

Who could benefit?

Small businesses without a dedicated security operations team

Freelancers and independent professionals

Students learning email security and phishing detection

IT support teams that need an accessible first-pass analysis tool

The long-term side-hustle opportunity is a lightweight security assistant for small teams that need understandable phishing triage without the cost or complexity of a full enterprise security platform.

Demo
Watch the demo: https://www.loom.com/share/ca8e3321c0274ec49f58d0d423f2813b

The demo shows the application running locally and demonstrates its phishing-analysis workflow.

Code
GitHub repository:
https://github.com/SM01010011/PhishBrief_AI

Explore the source code, setup instructions, and project implementation in the repository.

How I Built It
PhishBrief AI combines a Python application with local AI inference through Ollama.

Core technologies
Python — application logic and analysis workflow

Streamlit — interactive user interface

Ollama — local model runtime

Granite 3.2 Vision — the configured open-weight model used for local inference

The development setup also includes Gemma 3 1B as another locally available model.

The workflow is designed around analyzing user-provided suspicious content, identifying potential phishing indicators, explaining the associated risks, and presenting recommendations in a more understandable format.

The emphasis is on making AI-assisted security analysis accessible while exploring local inference as an alternative to sending potentially sensitive content to a third-party AI API.

This is an assistive analysis tool, not a guarantee that every phishing attempt will be detected. Users should still verify suspicious messages and links independently.

Why Does Open Innovation Matter?
Open innovation makes projects like PhishBrief AI possible for developers who do not have access to expensive infrastructure or enterprise security systems.

Open-weight models and tools such as Ollama make it possible to experiment with local inference, learn how AI systems behave, and build practical applications around them.

For this project, local inference offers three important advantages:

Privacy-conscious experimentation: Potentially sensitive email content can be processed locally, depending on the application's implementation and configuration.

Lower barriers to development: Developers can experiment without depending entirely on paid, hosted AI APIs.

Customisation and learning: Open-weight models make it easier to explore model choices, prompting strategies, and application workflows.

Open innovation does not automatically guarantee security or accuracy, but it gives independent developers more control over how AI-powered tools are built and evaluated.

My First User
My first target users for PhishBrief AI are individuals and small-business teams who need a simple way to understand suspicious emails without requiring advanced cybersecurity knowledge.

I'm looking for early testers who can try the tool with safe sample phishing emails, evaluate whether the risk explanations are understandable, and share feedback on its usefulness and ease of use.

The goal is to learn whether local AI-powered phishing analysis can make basic security triage more accessible to people without a dedicated security team.

I'm currently looking to gather feedback from early users and improve the project based on their experience.

Prize Categories
PhishBrief AI is my submission to the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle.

The project focuses on building a practical AI-powered tool around open-weight models and local inference.

Wildhunt_AI

Swarit Mathur — Sun, 11 Oct 2026 02:13:05 +0000

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

What I Built
WildHunt AI is a privacy-first, open-source AI scavenger hunt that turns your phone into a guide for exploring the physical world.

Instead of asking people to spend more time interacting with an AI chatbot, WildHunt gives them a reason to put the screen down, go outside, observe their surroundings, and come back only when they have something worth showing the AI.

The core loop is:

Choose a Hunt
↓
Get a mission
↓
Go outside
↓
Find the target
↓
Take a photo
↓
Local AI verifies it
↓
Unlock the next discovery

WildHunt supports different exploration styles such as:

🌿 Nature
🏙️ Urban Explorer
🎨 Colors
🪨 Texture
👀 Observation
📍 Landmark Hunt

The important part is that the AI isn't simply performing generic image classification.

It evaluates the submitted photo against the specific mission the user was given.

For example, if the mission is to find a naturally yellow object outdoors, the AI checks whether the photograph actually satisfies that mission.

Most AI products compete for your attention. WildHunt uses AI to give your attention back to the physical world.

Who is it for?

WildHunt is designed for anyone who wants a lightweight reason to explore:

-Students taking a break from studying
-Developers spending too much time at their desks
-Families looking for simple outdoor activities
-People exploring a new neighborhood
-Anyone who wants to turn a walk into a small adventure

Demo
🎥 Video Demo: https://drive.google.com/file/d/17hSBEzV7f4hfPKQCj0GlNb3MraOQQwBd/view?usp=sharing

The demo shows the complete WildHunt-AI experience:

1.Starting a hunt
2.Receiving a mission
3.Capturing a discovery
4.Sending the image to the local vision model
5.Receiving mission-specific verification
6.Continuing the hunt

The application runs as a responsive Progressive Web App, so it can be used directly from a phone browser without requiring a native mobile application.

Code
🔗 GitHub:https://github.com/SM01010011/WildHunt_AI

GitHub logo devanshshukla-3004 / WildHunt-AI
🌿 WildHunt AI
The screen gives you the mission. The real world gives you the answers.

Hacktoberfest 2026 Open Source AI React TypeScript Ollama License: MIT

WildHunt AI is a privacy-first outdoor scavenger-hunt PWA that uses local open-weight vision AI to turn real-world exploration into an interactive game.