What I Built

ListingCraft is a simple AI-powered tool that helps small sellers on Etsy, eBay, and Amazon turn rough product notes into polished, SEO-friendly listings.

If you've ever sold anything online, you know the pain: you have a great product but writing a compelling description feels like pulling teeth. You stare at a blank text box, type something generic, and wonder why nobody's clicking.

ListingCraft fixes that. Paste your rough notes — "handmade ceramic mug, blue glaze, 12oz, dishwasher safe" — pick your platform and tone, and get back a complete listing with a catchy title, engaging description, bullet-point features, and SEO tags. Done in seconds.

Who would pay for this? Any small seller who lists 10+ products a month. At even $5/month, it pays for itself after one listing that converts better. Etsy alone has 7.5 million active sellers.

Code

devkumar2313 / listingcraft

🏷️ ListingCraft

AI Product Listing Generator for Small Sellers

Turn rough product notes into polished, SEO-friendly marketplace listings — powered by open-weight AI.

✨ What It Does

Paste your rough product details → pick a platform (Etsy, eBay, Amazon) → get a polished listing with:

Catchy title
Compelling description
Key feature bullet points
SEO tags/keywords

🚀 Quick Start

# 1. Clone and install
git clone https://github.com/YOUR_USERNAME/listingcraft.git
cd listingcraft
pip install -r requirements.txt

# 2. Get a free Groq API key at https://console.groq.com/keys
export GROQ_API_KEY=your_key_here   # Linux/Mac
set GROQ_API_KEY=your_key_here      # Windows

# 3. Run
python app.py

Open http://localhost:7860 in your browser.

🤖 How It Works

ListingCraft uses Llama 3 (Meta's open-weight model) via Groq's free inference API. The model is fully open-weight — meaning anyone can inspect, modify, or self-host it.

Model: meta-llama/llama-4-scout-17b-16e-instruct (open-weight, Apache 2.0-like license)
Framework: Gradio (open-source UI library)
No data stored — your product…

View on GitHub

How I Built It

The stack is intentionally simple:

Model: Llama 3 by Meta — an open-weight LLM. I'm using the llama-4-scout-17b-16e-instruct variant, which is great at following instructions and generating natural-sounding copy.

Inference: Groq — free tier API that runs open-weight models. No API costs for a side hustle starting out.

UI: Gradio — open-source Python library for building web interfaces. The whole app is ~80 lines of Python.

Prompt engineering: A carefully tuned system prompt that acts as a marketplace copywriter, generating titles, descriptions, features, and tags tailored to the selected platform.

The entire thing is one Python file. No database, no auth, no complexity. Just the core value loop: paste details → get listing.

Why Does Open Innovation Matter?

Open-weight models make this side hustle actually viable. Here's why:

1. Cost structure that works. If I used GPT-4 or Claude, every listing would cost ~$0.01-0.05 in API fees. That doesn't sound like much, but for a $5/month tool serving hundreds of sellers generating thousands of listings, it adds up fast. Llama 3 via Groq's free tier? $0. When I scale up, I can self-host the model on a $50/month GPU server.

2. No vendor lock-in. If Groq changes their pricing tomorrow, I can switch to running the same Llama model on Replicate, Together AI, or my own hardware. The model is open — I'm not locked into anyone's ecosystem.

3. Customization potential. Down the road, I could fine-tune the model on actual high-converting Etsy listings. You can't fine-tune GPT-4. With open weights, I can build a model that's better at Etsy copy than any general-purpose API.

4. Transparency. My users can see exactly what model powers their listings. No black box, no surprise changes when the provider updates their model.

For a side hustle, open-source AI isn't just a nice-to-have — it's what makes the economics work.

DEV Community

ListingCraft — AI Product Listing Generator for Small Sellers