Building an AI Fashion Stylist

##Inspiration

### Thrifting
My friend who owns a thrift clothing store wanted to innovate and improve his customer's shopping experience. As opposed to traditional clothing stores that specialize in specific design and aesthetics, a thrift store offers enough variety and unpredictability to make the experience feel new everyday. Additionally, sometimes some of his stock comes defective, torn or faded due to their pre-loved nature. However, the idea came about to convert these clothes into custom tote bags. This ensures clothes are able to get a second and even third life of value.

###Sustainability
Our country Kenya banned plastic bags and mandates the use of environment friendly reusable bags, this tote project contributes to this vision of an eco friendly community with a bit more style.

### Vision
I also recently started watching a show called Gachiakuta, it is a new anime show with superpowered people called givers who are able to convert very dear items to them called Jinki into special instruments for various purposes. The main character called Rudo is special as he can convert other items into Jinki and draw out their true potential. This made me wonder how we can do something similar in the real world.

###Innovation
Additionally after attending Google devfest Nairobi I was able to learn how to use various models and google tools to accomplish previously arduous tasks as well as creative endeavours using nano banana and veo3.

### Product
All this culminated in this product called AI Fashion Stylist that allows us through the use of Artificial Intelligence, to make a positive impact on our environment via the styling, selling and upcycling clothes that would otherwise be considered unusable into creative designs using it's creative image generation capabilities.

### What it Does
AI Fashion Stylist provides AI-powered outfit suggestions based on your style goals, occasion, weather, and available wardrobe. Users can:

- Upload a garment image and describe their style context (e.g., "casual summer picnic," "office-appropriate," "date night")
- Get complete outfit recommendations with accessories and color palette suggestions
- View AI-generated visualizations of the styled look
- Download outfit recommendations as images or PDFs
- Access a curated inventory of styling items from Google Cloud Storage
- The app adapts to constraints (e.g., "office-appropriate," "carry-on only," "under 10 minutes to get ready") and generates tips for color harmony, silhouette balance, and ways to rewear pieces.

Additionally it allows for the designing of tote bags via upcycling of old garments. Users can:

- Upload garment images
- Select the design type eg plain, denim, khaki or corduroy
- View AI-generated visualizations of the styled tote bag design
- Download tote design recommendations as images or PDFs

### Challenges I Ran Into
- I initially wasn't able to publish the app to cloudrun as I was not conversant with the various google cloud services and thus had trouble configuring env variables. Additionally there were security considerations and cost management by making the app affordable to run and scale.

### Accomplishments that we're proud of
- Making an app that will have real tangible results environmentally and financially.
- Using AI for positive impact in the creative scene.
- Using google cloud services to deploy and manage an application successfully.
- Providing employment opportunities to the tailors who have managed to make the tote bags.

### What we Learned
- Learning Gen AI features and utilizing it's multimodal features to accomplish tasks.
- Learning google cloud services and deploying an application successfully.
- What's next for AI Fashion Stylist
- AI Fashion Stylist will be deployed to a second store to increase visibility and gain more user feedback.
- We will add more designs for upcycling to increase variety and ensure minimal wastage of material.
- Increase monetization of upcycled items to increase, train and empower more tailors.

### How I Built It
I built the first prototype using Google AI Studio using React google gen ai, google storage buckets and google cloud run. I then converted it into a NextJs app to increase security for my env variables and API keys. This move also allows the ability to add future curated experiences and addition of features.

Challenges I ran into
I initially wasn't able to publish the app to cloudrun as I was not conversant with the various google cloud services and thus had trouble configuring env variables. Additionally there were security considerations and cost management by making the app affordable to run and scale.

Accomplishments that we're proud of
- Making an app that will have real tangible results environmentally and financially.
-Using AI for positive impact in the creative scene.
-Using google cloud services to deploy and manage an application successfully.
-Providing employment opportunities to the tailors who have managed to make the tote bags.

### What we Learned
-Learning Gen AI features and utilizing it's multimodal features to accomplish tasks.
-Learning google cloud services and deploying an application successfully.

### What's next for AI Fashion Stylist
-AI Fashion Stylist will be deployed to a second store to increase visibility and gain more user feedback.
-We will add more designs for upcycling to increase variety and ensure minimal wastage of material.
-Increase monetization of upcycled items to increase, train and empower more tailors.


## Where to find it 

You can find the AI Fashion Stylist project [here](https://ai-fashion-stylist-711806383367.europe-west1.run.app/).

## Source code

You can find the source code [here](https://github.com/DixonKimandu/ai-fashion-stylist/tree/nextjs).

## Conclusion

AI Fashion Stylist is a project that allows you to generate fashion outfits based on the user's uploaded image and prompt, using Gemini API. It is a project that I built to help the environment and the community.