Artificial intelligence and fashion: a platform play
Artificial intelligence, often called AI or deep learning, will undoubtedly transform the fashion industry. This transformation will have a few important qualities: 1) AI will be a platform play, not something brands build and launch individually; 2) the effects will mostly be invisible to the consumer; and 3) AI is as much a mindset as it is pure technology. It will not be a quick fix — it will take time.
Before diving in, it’s important to understand what AI is.
This tweet from Christopher Mims is apt. Many articles in the press, as well as early companies in this space, are throwing around AI and other buzzwords, when in fact they are just building software. This is fine, but it’s important to be precise. AI is software written to learn and react without specifically coded parameters. Humans write an algorithm that learns, not one definitively defined. With AI, data is the key. Millions of data points train the algorithms so they continue to improve in a virtuous cycle.
AI is very hard. The biggest tech companies, from Google to Facebook to Microsoft are throwing an immense amount of resources towards AI, and there is still a ways to go. This has a crucial implication for the fashion industry: with a few exceptions, infusing AI into the industry will be a platform play. Individual brands simply do not have the resources or expertise to implement AI on the scale of its possibilities.
This means that AI will be the next big platform play for the industry, possibly mapping to the growth in ecommerce platforms. Yes, some brands built their own systems, but this will always be the exception. Ecommerce platforms, such as Demandware, Shopify and Magento succeed because of both the logic of a platform and the inherent limitations of a fashion brand’s technology skillset.
These platforms will exists both as APIs and full service applications. APIs that collect and process a brand’s data will be incredibly powerful. Brands will need to spend time and money turning their disparate data sets into a single firehose, which will feed the APIs. Whoever builds these APIs will own the plumbing of the next big platform shift in retail.
AI will mostly be invisible to the consumer
The most important uses for AI will be hidden from the consumer. Better aligning supply and demand will be AI’s most important development. Inventory is one of the most brutal aspects of selling physical things. Since there is still a ways to go until supply chains are agile enough to produce products on-demand, AI will provide advancements in the meantime. AI will affect everything from deciding how much of an item to order to determining where the finished goods should be shipped to maximize their sell-through. Even if a brand orders the theoretical right amount of products, distributing them to the right place is just as important. AI can help in both cases.
Other uses for AI include:
- Showing the right people the right products
- Working to minimize returns and shipping costs
- Learning about the sizing and fit of customers to design better fitting products
- Using pattern recognition to catch trends before they happen
- Building radically new shopping interfaces
A quick note about chat bots, which are much talked about now. New technologies need to be evaluated on one crucial axis: does it add value? Right now, many chat bot applications are making the experience worse for a customer. There are specific uses for chat bots, such as customer service and push notifications. But trying to turn everything into a chat interface is the wrong way to go about it. Problem, then solution, not the other way around.
AI is a mindset, and it will take time
Even though retail-based AI will be a platform play, brands will still have a ton of work to do. Fashion brands will be captive to platform risk, which is the baggage associated with adopting a mission-critical platform. The platform play means brands will have much less control because of their lack of technology expertise and resources. This places even more pressure on brands to tread lightly and carefully, from picking their partners to deciding how much they rely on outside vendors. Even so, brands will still need to add internal expertise to prepare their data for the platforms, constantly stay on top of advancements, and interpret the findings.
Most importantly, brands need to accept the right mindset in order to capitalize on AI. Millions of data points will better inform decisions, helping move the fashion industry away from problems induced from gut decisions, such as over and under buying. Adopting the right mindset is the biggest barrier for these brands to overcome, given their tepid adoption of technology so far. A pro-AI mindset is the only way to gain the in-house expertise needed to take advantage of all that is to come. If top executives don’t value AI, no one else will.
Originally published at loosethreads.xyz on April 10, 2016.