Artificial intelligence is the proverbial carrot dangling in front of any future-thinking business. Forecasting demand for bottled water to avoid stock shortages in the days leading up to a heatwave? Sure. Trained artificial intelligence can pull that off better than any human, exploiting under-used data that we might initially discount as out of hand.
Headline-grabbing promises – self-driving cars, smarter-than-human robots – aren’t really what AI is all about though. In our day-to-day lives, AI is simply a layer of intelligent technology applied to a process already in place. For example, forecasting supply chain demand, or enhancing the customer experience with autofill forms through image recognition. AI is being used to create something beneficial for end-users.
AI touches all sectors in 2020: for businesses, it’s creating waves of excitement but also ripples of trepidation. Meanwhile, savvy companies are keen to employ AI but are often unsure exactly how to go about it. For example, what’s the best way to invest to first convince colleagues of its value, and then how do you deploy it in ways that will have the most impact? Is it wise to spread investments across several projects or are more costly innovation hubs the way forward?
A model that offers a safer middle ground is the AI Factory. This brings together three separate entities: the group or brand that seeks to accelerate its data transformation; a Cloud provider; and the AI specialist, who can identify use-cases and deliver on them.
This triumvirate is the bedrock of the AI Factory. Each plays a crucial part because technology is no longer the barrier – rather, the challenge lies in successfully identifying the use-case and benefits of deploying AI and then delivering on it. Planning is the key element.
The concept relies on what we’ll call Future Teams: multidisciplinary teams combining AI professionals (what the provider specialises in) and business professionals (the client/brand). They bring together all the competencies required to successfully lead a project from A to Z: data scientists, data engineers, an ‘AI product owner’ who understands the requirements, and a DevOps engineer who ensures good supervision and long-term maintenance of the AI solution.
These teams, which combine business, engineering, and professional expertise, must work closely with the client. As such, it’s not a typical client/supplier relationship and requires lines of communications to remain open at all times.
The first objective should be the rapid development of a custom-made prototype, able to deliver added value quickly and efficiently. With close collaboration across the Future Teams, risks are better controlled and results more easily measured. A realistic timeframe to deliver operational solutions should be four to six months.
The AI Factory in action
Along with banks and insurance, retail offers vast quantities of data that can be leveraged from hundreds of thousands of products, transaction data, and loyalty cards. With all that information, the retailer should know its customers as well as Google, if not better.
This is particularly the case in e-commerce. Here it’s possible to monitor data from a shopper’s first click right up to order confirmation.
For example, French supermarket Carrefour employed the AI Factory methodology as a powerful growth accelerator. It first used AI to automatically predict the volume of home delivery orders based on customers’ previous buying habits – and was then able to predict daily sales volumes accurately to 94%.
It is now extending this forecasting approach to other elements of the business. This offers tangible business intelligence that allows it to operate more efficiently.
Businesses that employ AI across any sector put themselves in the best position to gain a competitive edge. However, success will be incremental. It means asking the right questions and escalating solutions once you know you are able to generate the right responses.