CraftContent.AI

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

What I Built

CraftCopy AI — Product Listings for Handmade Businesses

CraftCopy AI is an AI-powered writing assistant designed for independent makers and small handmade businesses.

Selling a handmade product involves more than making it. Sellers also need to write product descriptions, create social media captions, research searchable tags, and prepare listings for different platforms. For someone running a small shop alone, these repetitive tasks take time away from creating products and serving customers.

CraftCopy AI aims to simplify that process. A seller enters basic product information, such as the product name, materials, dimensions, price, target audience, and key features. The application generates editable product titles, descriptions, social media captions, search tags, and image alt text.

The initial audience is crochet artists, handmade jewellery makers, and independent online sellers.

The side-hustle opportunity is to help these sellers prepare marketing content more efficiently. Future versions could introduce batch listing generation, reusable brand presets, and specialized templates for different product categories.

Demo

Local application: CraftCopy AI currently runs locally.

Live demo: [Add your deployed application URL or a screen-recording link here.]

The demo will showcase the workflow from entering product details to generating and editing a product listing.

Code

GitHub repository: [Add your public GitHub repository URL here.]

The project includes a Python backend, a browser-based interface, model integration, a fallback generation workflow, and automated tests.

How I Built It

CraftCopy AI is built with Python, FastAPI, HTML, CSS, and JavaScript.

The AI integration uses Qwen2.5-0.5B-Instruct, an open-weight language model, through Hugging Face Transformers.

The intended workflow is straightforward:

1. The seller provides structured product details.
2. The backend prepares the input for the language model.
3. The model generates draft marketing content.
4. The seller reviews, edits, and copies the results.

The application also includes a template-based fallback when model inference is unavailable. This keeps the basic workflow accessible while distinguishing fallback output from genuine model-generated content.

The model can be downloaded and run locally, subject to hardware requirements and the model's license. This avoids requiring a paid proprietary model API for every generation request.

Why Does Open Innovation Matter?

I wanted to explore how open-weight AI could make a small business tool more accessible to independent developers and sellers.

A commercial writing tool built entirely around a proprietary API may introduce recurring inference costs, dependency on a third-party provider, and restrictions on how its underlying model can be customized.

Using an open-weight model provides a different approach.

Control over inference: The model can run locally, allowing users to keep product information on their own machines rather than sending every prompt to an external model provider.

Control over costs: Once the model has been downloaded, local inference does not incur a per-request fee from a third-party model API. Hardware, electricity, storage, and maintenance still have costs.

Freedom to experiment: Developers can evaluate different compatible models, adjust prompts, and explore specialized workflows for handmade product categories.

Accessibility: The project can be cloned, inspected, modified, and extended by other developers.

Local operation does require an initial model download and sufficient computing resources. The goal is not to claim that open models are automatically better at every task, but to make experimentation and deployment more accessible without making a proprietary API a mandatory dependency.

For a small side-hustle product, that flexibility matters: it creates room to experiment with different approaches without tying every generation to someone else's pricing and service availability.

My Agent Session

Optional. I will add a DevRelay session if I capture one that demonstrates the development process.

My First User

CraftCopy AI is intended for independent makers and small handmade businesses. The next milestone is to test it with a real seller and collect feedback on listing quality, editing effort, and missing features.

I have not yet established whether a seller would pay for the product, so validating that demand is part of the next stage.

Prize Categories

I will select the applicable partner categories according to the official challenge rules before publishing.