我打造了 HustleKit AI：為個人網店賣家而設的 AI 增長助手

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

What I Built

Running a small online business means wearing too many hats. You need to write product descriptions, create marketing content, understand sales, and figure out what to do next—all while trying to grow your business.

I built HustleKit AI, an AI-powered workspace designed to help solo online sellers, indie makers, and creator-led businesses handle these tasks more efficiently.

Instead of treating AI as just a text-generation chatbot, HustleKit AI combines content creation with sales analytics and structured growth recommendations.

Here's what it can do:

• Product Studio: Generate product marketing copy and structured content using AI.
• Sales Insights: Analyze sales records, compare performance across channels, and surface useful business metrics.
• AI Growth Advisor: Turn sales data into actionable recommendations through a multi-step agent workflow.
• Business Dashboard: Bring key metrics and tools into one workspace.
• Persistent Settings: Store business preferences and profile information between sessions.

The idea is simple: give independent sellers a practical starting point for both marketing their products and deciding how to grow.

Demo

🚀 Try HustleKit AI: https://hustlekit-ai.onrender.com

The live application connects to a production backend, MongoDB Atlas, and a hosted Gemma model.

Code

💻 GitHub Repository: https://github.com/Pranjal-debug/hustlekit-ai

The repository contains the application code, setup instructions, environment-variable templates, and implementation details.

How I Built It

HustleKit AI uses a full-stack TypeScript architecture.

Frontend

• React, Vite, and TypeScript
• Tailwind CSS for styling
• Lucide icons and a responsive dashboard interface

Backend and data

• Node.js and Express
• MongoDB Atlas for persistent business data
• API routes for content generation, analytics, settings, and growth recommendations

AI and agent workflow

• Gemma 3 12B IT for generating marketing content and growth recommendations, accessed through OpenRouter.
• Mastra to orchestrate the Sales Growth Advisor workflow.
• Zod to validate structured AI output before it is accepted and persisted.

The most interesting part is the Growth Advisor.

Rather than asking a model to invent business advice from a vague prompt, the workflow separates the process into four steps:

1. Load business data from the database.
2. Build evidence from sales metrics and business context.
3. Generate recommendations using the AI model.
4. Validate the output against a schema before saving the report.

This separation makes the workflow easier to reason about and helps ground recommendations in actual business metrics.

I also built and deployed the application, configured the production environment, and verified the main application flows. The final release build succeeded, and the project's backend test suite passed 49 tests across six suites.

Why Does Open Innovation Matter?

Small businesses should not need a large engineering team or an expensive enterprise software stack just to experiment with AI-assisted workflows.

Open-weight models such as Gemma give developers more choice in how they build AI-powered products. They make it possible to experiment with model capabilities, switch inference providers, and design workflows around a model instead of locking the entire application to a single proprietary ecosystem.

For HustleKit AI, Gemma provides the language-generation capability, while Mastra lets me build a more structured workflow around it. MongoDB Atlas supplies persistent data storage.

I'm using hosted inference through OpenRouter rather than running the model locally, but the application is built around an open-weight model and a modular architecture.

For me, open innovation is about making useful AI capabilities more accessible while giving developers the flexibility to understand, adapt, and extend the systems they build.

Prize Categories

• Gemma: Use of Gemma 3 12B IT for AI-powered marketing content and growth recommendations.
• Mastra: A multi-step Sales Growth Advisor workflow with evidence gathering and schema validation.
• MongoDB Atlas: Persistent storage for sales records, business settings, and generated advisor reports.

What I Learned

Building HustleKit AI taught me that integrating an AI model is only one part of building a useful AI product.

The harder engineering challenge is connecting model output to real application data, validating results, handling failures, and presenting everything in an interface that people can actually use.

I also learned the importance of treating deployment and testing as part of the product—not as tasks to leave until the end.

HustleKit AI is my attempt to turn those lessons into a practical tool for people building businesses on their own.

Thanks for checking it out!