Landing Page of Fankaar.ai
Type it out, or just speak to us!
Features we offer
Our Generated Marketing Suite
Landing Page
Generated Comics

About the Project

Inspiration

We were inspired by the everyday struggles of local artisans and small creative businesses.
They are incredible at their craft, but when it comes to digital marketing, multilingual content creation, or online sales, the process becomes overwhelming.
We wanted to build a tool that lets artisans focus on creating, while AI takes care of the digital hustle.

How We Used Kiro

To speed up development and maintain clean, modular code, we integrated Kiro, an efficient coding assistant tool, into our workflow.

It helped us auto-generate boilerplate code for FastAPI routers, database models, and React components, saving time on repetitive tasks.

Kiro’s context-aware suggestions ensured that our services followed consistent design patterns, reducing bugs caused by mismatched interfaces.

We used it to document APIs and data flows on the go, making collaboration smoother across frontend and backend teams.

By offloading low-level scaffolding to Kiro, we could focus more on core features, ML integrations, and user experience design.

In short, Kiro acted as our developer co-pilot, ensuring both speed and maintainability while building a multi-service AI-powered system.

What We Learned

How to design AI for inclusivity, especially with multilingual and culturally-aware content generation.
The importance of building intuitive UIs that even first-time digital users can navigate.
How to connect multiple services — FastAPI, React, ML models — into one cohesive system.
Applying machine learning models (ResNet50 + CBM) for both classification and price recommendation.
The value of microservice architecture for modularity, independent scaling, and easier debugging.

How We Built It

Frontend: React with modern UI/UX design and multilingual support.
Backend: FastAPI with modular routers for content generation, classification, inventory management, and social media automation.
Microservices: Each service (content generation, classification, pricing, social media, storage) runs as an independent microservice, allowing for easy scalability and fault isolation.
AI/ML:
- Gemini + Vertex AI for natural language generation.
- ResNet50 for image classification and craft recognition.
- CBM (Class Balanced Metrics) for accurate pricing suggestions.
Database: SQLite for storage and persistence.
Services Integration: Google TTS, Veo, FFmpeg MCP, and Nano Banana for rich media generation.

The workflow:

Artisan uploads an image.
AI generates descriptions in multiple languages, recommends pricing, and classifies the craft.
Inventory service + holiday data suggest what to make next.
Social media service auto-generates banners, thumbnails, and marketing posts.

Challenges We Faced

Ensuring AI-generated text preserved authentic artisan voice while being professional.
Handling multilingual accuracy and cultural nuances.
Integrating multiple ML and AI services without compromising performance.
Managing scalability — achieved using microservice architecture, which allows us to independently scale services like content generation or media creation without bottlenecks.