This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
What I Built
ReSellReady is a focused AI listing agent for side-hustle sellers. A seller enters a product name, marketplace, writing voice, and the facts they know. ReSellReady adapts its instructions to the product category, then generates an editable marketplace title, description, tags, buyer checklist, and listing-health score.
It supports vintage, handmade, food, print-on-demand, and other independent-seller products. A freelancer, Etsy seller, vintage reseller, home baker, or independent maker could pay for it to turn rough product notes into a stronger listing without hiring a copywriter for every item.
The agent is deliberately narrow: it routes category-specific questions, follows safety constraints, avoids inventing product facts, identifies missing information in the checklist, and returns a structured listing rather than pretending to be a general chatbot.
Demo
Live demo: https://resellready.onrender.com
Health check: https://resellready.onrender.com/api/health
Code
Repository: https://github.com/ayushmankoley/Resellready
How I Built It
ReSellReady uses Google's open-weight Gemma models:
- Primary model:
gemma-4-31b-it - Backup model:
gemma-4-26b-a4b-it
The browser UI sends listing details to a small Node.js backend. The backend builds a category-aware agent prompt containing the seller's facts, marketplace, tone, safety rules, output contract, and an example. It calls Gemma server-side, validates the returned JSON, recovers JSON wrapped in code fences or explanatory text, and retries with the backup model for temporary provider failures or malformed model output.
The frontend includes product adaptation, photo previews, a loading state, hosted-model labeling, error fallback, and copy-to-clipboard actions. Render hosts the public Node.js service, while the Google API key remains a server-side environment secret.
Why Does Open Innovation Matter?
Open innovation makes this product easier to adapt to the real vocabulary of small sellers. A vintage reseller, home baker, or handmade maker has different facts and safety concerns; the prompt and model can be tuned around those workflows instead of forcing every seller into a generic copywriting experience.
Using an open-weight Gemma model also reduces dependence on a closed model vendor's product decisions and makes the agent's behavior inspectable and changeable. The model can be swapped, backed up, evaluated, or eventually fine-tuned for a niche category. Hosted inference keeps the demo accessible while the application still keeps the provider key out of the browser.
Prize Categories
Best Use of Gemma. — Our open source model
Best Use of Render — Render hosts the public ReSellReady frontend and backend as a single Node.js web service, with a public live demo and /api/health deployment check.
Best Use of GitHub Copilot — GitHub Copilot was used throughout the build to develop, debug, refine, and ship the ReSellReady agent.
Top comments (0)