HuslteAI：面向電商商品刊登的開源 AI 助理

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

HuslteAI: Your Local AI Assistant for E-Commerce Listings

What I Built

I built HuslteAI, a local AI-powered product listing assistant designed for independent sellers, sma

HuslteAI turns this process into a simple workflow. Enter a product name, describe its features, choose a marketplace and writing style, and generate:

• Product titles that clearly present the item.
• Product descriptions tailored to the selected marketplace.
• SEO keywo rds to help sellers optimize their listings.
• Social media captions with relevant hashtags.

The goal is to make useful AI writing tools accessible to people who cannot justify paying for multiple AI subscriptions.

Demo

🚀 Try HuslteAI: https://aryansinghsuryavanshi28.github.io/HuslteAI/

The frontend is hosted on GitHub Pages. To generate listings, the app is designed to connect to a locally running Ollama instance on the user's computer.

Code

💻 GitHub Repository: https://github.com/AryansinghSuryavanshi28/HuslteAI

The project is built with HTML, CSS, and JavaScript, with Ollama providing the local model interface.

How I Built It

I built HuslteAI using standard web technologies and an open-weight language model running locally.

The core stack is:

• HTML for the interface and product input form.
• CSS for the responsive design and listing studio.
• JavaScript for prompt construction, local model requests, result parsing, copying, and downloading generated listings.
• Ollama to run and serve the language model locally.
• Qwen2.5 3B (qwen2.5:3b) as the open-weight model that generates the listing content.

When a seller submits a product, the JavaScript code creates a prompt containing the product details, selected platform, and desired writing style. It sends that prompt to Ollama's local generation endpoint at http://localhost:11434/api/generate.

The model is instructed to return four sections: [TITLE], [DESCRIPTION], [KEYWORDS], and [CAPTION]. The app parses the response into separate result cards and allows the seller to copy or download the output.

I also included instructions to avoid inventing product features, certifications, discounts, reviews, health benefits, or guarantees. Sellers should still verify all generated content before publishing.

Why Does Open Innovation Matter?

For this project, open innovation is about making AI practical, affordable, and more controllable for small businesses.

1. Local inference and privacy

Instead of sending product details to a hosted text-generation API, HuslteAI is designed to send them to an Ollama instance running on the seller's own computer. This helps keep product notes and business information under the user's control.

2. Lower recurring costs

Once the model is downloaded, local generation does not require a per-request fee from a hosted AI provider. Users still need suitable hardware and electricity, and the initial model download requires internet access.

3. Freedom to experiment

The model can be replaced with another model supported by Ollama. That makes it possible to experiment with different model sizes, writing styles, and product-specific prompts without rebuilding the entire interface.

4. Less dependence on a single provider

The project does not depend on a paid, proprietary text-generation API. This gives the developer more flexibility to change models and adapt the tool as requirements evolve.

5. A realistic side-hustle opportunity

Small sellers need repeatable tools that save time. A local listing assistant could grow into a paid product through features such as saved brand voices, batch listing generation, marketplace-specific templates, and reusable product catalogs.

Open innovation does not automatically make a model better than every closed alternative. For HuslteAI, its main advantage is the opportunity to build a more controllable, locally run, and potentially low-cost workflow.

My Agent Session

Not included in this submission.

Prize Categories

I have not selected a partner prize category yet. I will add the applicable category after checking the challenge's official prize-category requirements.

Built independently by @AryansinghSuryavanshi28.