Fashion Recommendation

With the advent of Generative AI, Search and finding a product is being revolutionised. We are
moving away from a single open text box experience to something more conversational and this
will enable product discovery and recommendations to be a lot more powerful than they are
today by way of being able to truly understand the user’s needs in a more human-like
conversational way. Fashion is one of the categories where discovery will get reset.
As part of this challenge, teams need to create an Gen AI-powered fashion outfit generator for
Flipkart that revolutionizes the way users discover and create personalized fashion outfits, in a
natural conversational way. The outfit generator should leverage user's past purchase history,
preferences based on browsing data, and insights from social media trends to offer tailored and
on-trend outfit recommendations.
The fashion outfit generator should have the ability to analyze a user's past purchase history and
understand their preferred style, color choices, and favorite brands. By considering these
preferences, the generator should look to suggest outfits that align with the user's unique fashion
taste. Additonally, the generator can take into account the types of clothing items the user
frequently views or adds to their cart, ensuring that the outfit recommendations are relevant and
appealing.
In addition to individual user data, the fashion outfit generator should tap into social media
trends to provide up-to-date fashion recommendations. It should be able to analyze current
fashion trends, styles, and influencers on platforms like Instagram, Pinterest, and fashion
blogs. By combining this data with the user's preferences, the generator can suggest outfits
that are not only personalized but also in line with the latest fashion trends.
The generated outfit recommendations should be complete and well-coordinated, including
clothing, accessories, and footwear etc. The generator should consider factors such as the user's
body type, occasion (e.g., casual, formal, party), and regional and age preferences (Ex. Young 20
year old woman looking for a Diwali outfit in Mumbai should be different to 35 year old woman in
Muzzafarpur looking for a Karwa Chauth outfit) to offer appropriate and versatile outfit
suggestions. Users should also be able to interact with the outfit generator to give it feedback in
terms of what they like, dont like and be able to tweak the outfits in the manner of a conversation
(Ex. I like the top, but the jhumkas are boring, give me something else).
The ultimate goal of the fashion outfit generator is to enhance the user's shopping experience on
Flipkart by providing them with personalized, trendy, and cohesive outfit ideas. Users should feel
inspired and confident in their fashion choices, knowing that the generator has considered their
preferences, browsing habits, and the latest fashion trends.
ChatGPT

{
  "name": "gpt-4",
  "config": {
    "description": "The GPT-4 model was built with broad general knowledge and domain expertise. ",
    "provider": "openai",
    "family": "gpt-4",
    "features": [
      "chat",
      "functions"
    ],
    "region": "us",
    "availableRegions": [
      "us"
    ],
    "maxTokens": 8192,
    "maxInputTokens": 6144,
    "maxOutputTokens": 2048,
    "pricing": {
      "tokenRatio": 3.3333
    },
    "interactionMaxMessages": 100,
    "visible": true,
    "deprecated": false,
    "temperature": 0.7,
    "frequencyPenalty": 0,
    "presencePenalty": 0,
    "tags": []
  }
}