Shaape.ai

Where Fashion Meets AI

Comment

Inspiration

Inaccurate Fit Recommendations

With 67% of online shoppers reporting dissatisfaction with fit, an AI-driven solution can reduce return rates by up to 30% through personalized sizing recommendations. [1]

Price Volatility

Price fluctuations can reach 20% in a single day, leading to consumer distrust; an AI tool that tracks and predicts pricing trends can enhance customer loyalty by providing price stability. [4]

Reducing Carbon Footprint

The fashion industry alone generates 92 million tonnes of waste annually, with returns contributing to this figure. [5]

Reverse Logistics

Return rates in online fashion can soar as high as 40%, but leveraging AI for virtual fitting rooms can decrease this by 25%, saving retailers millions annually. [3]

Inefficient Deal Comparison

Consumers waste an average of 7 hours per month searching for the best deals, but an AI extension can streamline this process, potentially increasing conversion rates by 25%. [2]

What it does

Shaape.ai aims to enhance the fashion shopping experience through a dual-platform approach. As a Chrome extension, it provides users with Fit & Occasion Recommender, Dynamic Price History Analytics, VibeMatch ratings, and Fashion GPT. Simultaneously, as a Shopify plugin, it empowers businesses to offer these AI-driven features (Powered by GPT 3.5) to their customers, simplifying the shopping process and boosting satisfaction.

Features

Fashion GPT: Intelligent Search Assistant

Revolutionize product discovery with our advanced semantic search, delivering precise and relevant results fast.

Price Insight: Dynamic Price History Analytics

Unlock savings with real-time price trends and analytics, guiding you to the best buy moments.

Fit & Occasion Recommender: Smart Sizing & Event Matching

Perfect your fit and event attire with our smart sizing and occasion matcher, driven by cutting-edge algorithms, NLP, and GPT 3.5.

VibeMatch: Personalized Fashion Suitability Score

Discover your perfect style with VibeMatch, providing tailored fashion scores based on your unique data.

How we built it

We leverage a sophisticated composite model integrating multiple recommendation engines with specific parameters and mutual dependencies. Utilizing intent generation and advanced semantic search, our approach simplifies production debugging and ensures rapid, accurate results at scale, delivering exceptional performance and user satisfaction.

Privacy Handling:

  • AES (Advanced Encryption Standard)
  • Data Masking
  • AWS KMS
  • End-to-End Encryption (E2EE)

Open Source Tools:

  • TiDB - Database enabled with vector semantic search
  • FAISS - Library for efficient similarity search and clustering of dense vectors
  • Langchain/NLTK
  • Google BERT - For custom data training and recommendation
  • Redis - In-memory data structure store
  • Telemetry - 24/7 Monitoring system with high availability

Challenges we ran into

  • Finding the best composite model approach for our Intent Generation Service, that accurately index our data, also normalise multiple areas like fittings, colours especially.
  • Working with the Prompt Engineering, to fine tune not only based on garments and their sizes and fitings but with the custom properties we generate using our Intent service

Accomplishments that we're proud of

  • Unique problem statement: Businesses that have implemented AI shopping assistants have reported a 30% increase in sales. This can be attributed to the personalised recommendations and improved customer engagement that our product aims to offer.
  • Brainstorming and the idea to understand how we can make the fashion purchases more seamless my making the descision making process for the end consumer more smoother and accurate.

What we learned

FAISS with TiDB Vector Search, can create a powerful system capable of handling large-scale similarity search tasks with high performance and accuracy. BERT is a really powerful language representation model that has been a big milestone in the field of NLP, which helped us sort our intention service and product classification processes.

What's next for Shaape.ai

  • Virtual Try-Ons: Facilitate virtual try-ons with AI styling suggestions for a personalized shopping experience.
  • Curated Lookbooks: Create curated lookbooks to help users complete outfits effortlessly.
  • Image Recognition and Suggestions: Implement AI-driven image recognition to suggest similar products based on user-uploaded images
  • On-Demand Notifications: Enable users to set up notifications via WhatsApp and other channels for price drops and new arrivals.

Steps to run the extension

Built With

Share this project:

Updates