Trent Gillespie says the two biggest mistakes companies make when introducing artificial intelligence is adding AI to existing processes rather than rebuilding processes around AI, and not getting senior executives aligned on their AI goals and priorities.
Drawing on his experience as a senior tech executive at Amazon.com Inc., Gillespie has developed what he calls his AI Sprint framework to help companies apply AI effectively to address their needs—and generate ROI as a result.
Gillespie, now CEO of consulting firm Stellis AI that helps companies deploy AI, will outline his approach in a keynote address entitled “Designing an AI-enabled business” at Distribution Strategy Group’s Applied AI for Distributors conference next month in Chicago.
At the heart of his AI strategy is emulating Amazon’s determination to make AI central to its business. “Amazon realized it can’t scale to support the world with humans,” Gillespie says. “There are too many transactions, too many products, too many customers. Instead, they decided they had to be AI-first.”
How is that approach different from how companies are now deploying AI? Gillespie gives the example of reconciling invoices with purchase orders, a common task for distributors.
Today, an employee may receive an invoice, enter it into the payables system, and manually reconcile it with what was ordered. Too many companies will add an AI product designed to provide the employee with more information, and they may see some improvements as a result, Gillespie says. But, he says, to get maximum impact they should redesign the process so that AI does all the routine work and only escalates problems it can’t handle to a human.
“You’re not adding AI on top of the process. Ideally, AI is doing all those steps and, if there are issues, that’s when you get humans involved.”
In companies deploying AI effectively, Gillespie says, senior leadership is aligned on how fast to move and what to invest in. “Ultimately,” he says, “the succeeding companies are one where the leadership team is on the same page and has created some kind of organizational plan. At a minimum, they’ve got someone who is responsible.” Gillespie says it’s surprising how often no individual is put in charge of AI projects, given their significance.
One other misconception he often sees is the idea that a company must get all its data into perfect shape before it can get started with AI.
Gillespie agrees that AI requires good data, and that many companies have work to do to get their data into shape. But he says his advice on data-cleansing is different from what companies often hear.
“Most vendors say that to make AI work you have to have perfect data, and companies wind up spending months getting their data perfect,” Gillespie says. “I tell businesses something different: You need to get the data correct for the top use case where you’re using AI.
“Figure out the best use cases and get the data to be good for those. You don’t need to do it all at once.”
That should be reassuring to distributors who often have mountains of data in varying formats and residing in many siloes within the company’s network. Gillespie promises to outline an approach at the DSG conference that focuses on the best AI opportunities—and that produces positive results.
Don’t miss any content from Distribution Strategy Group. Join our list.
Don Davis, former editor-in-chief of Internet Retailer magazine and Vertical Web Media, is a freelance writer based in Chicago. His experience in retail and distribution goes back to his childhood when he worked in the toy wholesale business founded by his father and two uncles and in their discount department stores located throughout the New York metropolitan area.