What’s an AI? It's a machine having human like cognition, dexterity or intelligence. So, like cars which can drive self, there can be apparels which can recommend themselves like a personal shopper would have.

A choice of hiring a personal shopper can help you shop quicker and way more cool apparels, simply because of 1:1 attention & advice. How much ever desirable but as an option, it is accessible to only a few. Let me take you through how one can put together some cross-domain tech to enable the same through AI.

The Turing test equivalent for AI based personal shopper would be a system which can handle queries like - ‘elegant dresses for office party that can make me look slimmer’, ‘does this dress suit my dusky skin tone, triangle body shape or heavy upper body’, ‘show me dresses of my kind’.

Let us examine the intellection carried out by a personal shopper while answering above queries & how AI can match the same:

1) Know your user

A stylist observes its users to learn about their persona and broad taste preferences, e.g. ‘she likes subtle prints with darker colors & overall sporty attire’.

AI for it:

Recently there is no dearth of users’ interaction data - click streams, social media comments, uploads. The Machine learned systems that consume enormous data and synthesize what users care or like most are already in practice. Google, Bing, Facebook & plethora of other consumer data companies rely critically on such techniques to target ads or content.

2) Handle trial room dilemmas

A good stylist or a close friend helps customers in addressing their usual dilemmas post trial room e.g. ‘does it work for my round tummy’, ‘does it complement my skin tone’, ‘does it work for my narrow shoulders’ etc. It is crucial for purchase decision.

AI for it:

Systems are increasingly able to model the human knowledge as artificial learning. All of the styling theory available as textbooks can be crawled, mined and synthesized to work as a knowledge graph using Artificial Neural Network. With such a knowledge graph, it is possible to generate human like suitability advice around an apparel individually for every user. There are firms which are already doing it.

3) Handle users’ natural intents

A stylist has to address the thematic and aspirational user queries e.g. ‘bohemian dress for beachwear’, ‘glamourous dress for petite’, ‘casual shirts for office for short height’ etc.

AI for it:

With advances in deep learning and computer vision, churning visual data and generating human like segregation/intelligence based on occasion, style statement and body parameters is possible now. Such visual intelligence coupled with recommendation engine enables the system to query the best products for what users naturally ask.

4) Personal connect

A stylist has to speak to users directly rather than getting a form filled.

AI for it:

Conversational interfaces using Natural Language Processing are already a reality that can chat with users in natural language as good as a person.

We are standing at the brink of AI revolution. The technology is almost there to start it off. Personal styling, a job best done by a human and a highly sought after but affordable to few experience, is one of the many areas where AI wows to replace humans.

By getting Natural Language Processing, Deep Learning, Styling Theory, and Recommendation Engine to work in tandem, it is indeed possible to have machines that can style people.


Editor's Note:
It's enchanting to read about AI in Fashion, from a CEO himself. Team TowardsAI is totally excited to learn more about AI in its various forms, and we look forward to read your Comments, for the same. Comment. Share. Subscribe. Thank you. :-)

Komal Prajapati

Komal is the CEO of Fabulyst, an AI startup in Fashion! They received their 1st round of investment from iBHubs (ibhubs.co) .