Kalapana-AI

Landing Page
Auth
Artisan Dashboard
Add Product(Dynamic Pricing + Enhanced Images)
Add Product(Storytelling Images)
Add Product(Craft DNA)
Muse Results
Artisan Mentor

About the Project

Inspiration

KalpanaAI was inspired by a gap I observed between India’s rich artisan heritage and the digital economy.
Millions of artisans create culturally valuable and sustainable products, yet remain invisible online due to the lack of storytelling, fair pricing, and access to global markets.
I wanted to build a platform where AI does not replace human creativity, but amplifies it, preserving cultural identity while enabling fair and scalable commerce.

What I Learned

Through KalpanaAI, I learned how to design responsible, production-grade AI systems instead of isolated ML demos.
Key learnings include:

Building multi-agent AI architectures on Google Cloud
Using Retrieval-Augmented Generation (RAG) to eliminate hallucinations
Designing voice-first, multilingual AI experiences for non-technical users
Balancing scalability, cost efficiency, and ethical AI design

This project significantly improved my system design, cloud engineering, and product thinking skills.

How I Built It

KalpanaAI is a serverless, agent-based AI platform built entirely on Google Cloud.
Each capability is handled by an independent AI agent deployed on Cloud Run and orchestrated via a central workflow.

Core technologies:

Vertex AI (Gemini & Imagen) for text, vision, and image generation
Firestore-based RAG to ensure cultural accuracy
BigQuery for dynamic pricing and demand forecasting
Next.js + TypeScript for a fast, mobile-first frontend
Firebase Auth & Cloud Storage for secure data management

A single photo and voice input is transformed into a complete marketing and commerce kit in under 60 seconds.

Challenges Faced

The biggest challenge was preventing AI hallucinations in cultural content.
Generic generative models often produce inaccurate or insensitive outputs.
I solved this by building a curated craft knowledge base and enforcing strict RAG pipelines.

Other challenges included: