top of page
  • Jade Woo

AI Shopping Assistants

I am chronically indecisive- a perpetual victim of analysis paralysis. Combine that with a proclivity for shopping, I tend to either over-buy or pester my friends for their opinions. Though I will probably never stop asking for other peopleʼs advice, I would love to have the opportunity to consult a stylist, to tell me if a top fits my personal style or if I am being influenced by the latest trends.


When I shop, I am always put in a situation where I am given too many things to look at. For example, as I am scrolling through a website, looking for the perfect everyday top that is more sophisticated than a t-shirt but more casual than a blouse, I am overwhelmed with all the different designers, and the PAGES ON PAGES ON PAGES of different tops  If only there was a way I could describe what I am looking for so that the webpage could present me with 5 options that match my needs.



What are AI Shopping Assistants?


With the rise of Generative AI, every industry is chomping at the bit to figure out how to best utilize it. For e-commerce, a trend seems to be AI Shopping Assistants.

We are all familiar with that tiny little customer service chat bot that appears at the bottom right of the screen when visiting a retail site, but we never engage with it for “shoppingˮ purposes, unless we need to make a return.

Today, AI shopping assistants can tell customers what to wear to a Hawaii wedding in April, how to dress like Lenny Kravitz, and more. Retailers and e- commerce companies are starting to implement generative AI shopping assistants as a way to help customers explore their product catalog in a more conversational and targeted way- it feels like talking to someone about what to wear.

In an ideal world, these assistants would help me find the best product for any given situation. The main benefits of AI Shopping Assistants are:


  • Reduce time spent scrolling through a website

  • Provide catered results specific to my style, event needs, weather, etc.

  • Discover new styles, brands, and outfit combinations

  • However, even with these benefits, todayʼs AI assistants are far from ideal.



What are the downsides to AI Shopping Assistants?


Personal aesthetic is not taken into account

Much of shopping is trying to find pieces that fit my overall personal style. If, on a multi-brand website, I ask for a casual outfit for my tech job in New York, I get a ton of results - but I wouldn't wear any of it. They do not fit my personal aesthetic or style. It is an important element that these AI assistants are missing. I can provide more information about the brands I like, but it still does not seem to yield the right results. Even when data on previous purchases is available, the assistants I have tried do not seem to take them into account when suggesting new clothing items. It seems like a waste to not use that sort of data, passing on the opportunity to extrapolate some of my preferred colors, styles, silhouettes, and brands from my previous purchases.


Body type/shape are not taken into account

There is a recent trend where people have started paying more attention to how clothes can look different depending on body type. Every personʼs body is different and that means clothes can look different depending on their body type. Obviously, it comes down to personal opinion and style, but I would like to know how an item looks on people with my body type (short torso and longer legs) or see what certain clothes look like on people with different body types. This would help me fully envision how something might actually look like on me, instead of having to buy it first. Since Generative AI can create images from descriptions, it seems plausible to one day have AI shopping assistant show me how a specific pair of jeans would look on someone with long legs and a short torso.



Clunky experience overall

While testing several AI shopping assistants, I often felt they kept asking me the same questions about my preferences without actually taking them into account. The response was only updated to include the words I used but the search results would pretty much remain the same.



Ideas on how we could improve on these AI Shopping Assistants


The biggest challenge AI Shopping Assistants face is giving customers personalized results based on their aesthetic. AI assistants are informed by self-reported data, and it is difficult to describe an aesthetic

Here are some some thoughts on how to solve that:


Leverage customerʼs past purchases

Why? This can give the AI a basic idea of styles, silhouettes, colors, and fit that a customer might gravitate to. Previous purchases and returns are a good indication of what has and has not worked in the past, and it should certainly be used to inform future purchases.


Create a common vocabulary that the everyday customer can use to describe how they want their clothes to work for them

Why? I am not entirely sure if thereʼs a way to describe how clothes should fit the body. I know how to describe the way I do not like clothes to fit me: I do not want a top to show my mid-drift; I do not like the tight feeling around my quads when I wear certain skinny jeans. But I do not know if I have the words for when something fits me well, other than “I like how this looksˮ. Maybe this is a personal problem, but if there was a way these AI shopping assistants could help us find the words to get to our ideal fit, that would be very helpful.


Perhaps prompt transformation techniques, which modifies usersʼ prompts to get more accurate results, can be of use here. If AI shopping assistant’s  were to adjust the wording of my prompts to make it more specific to my shopping preferences, maybe it would yield better results.


Themes to clothes

Why? Shoppers are labeling different categories of aesthetics, to direct them to find certain things. Brands or e-commerce sites can help distinguish themselves by showing shoppers their take on these categorized aesthetics using their own merchandise, through their own AI shopping assistants.



Closing remarks


In the (hopefully) near future, retail sites will look like my Instagram feed, showing me items curated specifically for me. AI shopping assistants have the potential to be incredibly personalized and become the friend, stylist, and retail worker that everyone needs. I am excited to see if more e-commerce sites start adopting AI shopping assistants, or if they will utilize Generative AI more behind the scenes.



What are your thoughts on AI Shopping Assistants? What do you think are the advantages and/or possible disadvantages of fully personalized shopping tips? Let me know at hello-nyc@bontouch.com

13 views0 comments

Commenti


bottom of page