The concept of artificial intelligence has been around for decades, but distribution company decision-makers could be forgiven if it still seemed foreign or far-off.
In the months since the rollout of ChatGPT and other so-called “generative AI” systems, however, the AI revolution has emerged in earnest — and made it increasingly imperative that businesses of all types get on board.
That prospect can seem unfamiliar and daunting, but leading technology vendors in the business-to-business sector recently offered words of encouragement — as well as warning — to attendees at the Industrial Supply Association’s annual convention earlier this spring.
AI, they said during a panel discussion at ISA24 in Charlotte, North Carolina, isn’t as difficult to get started on as it might seem; the real danger, in fact, lies in hesitating — and ending up left behind.
Shifting the Burden
For distributors worried that investing in AI will require enormous investments in technology, manpower or infrastructure, the four panelists at the ISA conference — Proton CEO Benj Cohen, PROS Vice President Greg Davoll, Verusen founder Paul Noble and Continuum founder Alex Witcpalek — said that partnering with software-as-a-service companies, such as theirs, would alleviate most of those concerns.
Panel discussion on "Navigating AI in the Industrial Channel" at ISA24, Charlotte, N.C., April 17, 2024.Andy Szal/Industrial Distribution
Rather than making their businesses more cumbersome, Witcpalek said, the entire point of AI is to make things easier — from cleaning up inconsistent or faulty internal data to improving “manual” or “arduous” tasks.
“If you partner with the right partners, they’ll help you with your data problems,” Witcpalek said.
Davoll said that PROS “rarely” runs into companies that are interested in hosting AI capabilities on their own; instead, the company rolls those technology demands and server requirements into its own costs.
‘A Good Place to Start’
The good news for distributors, the panelists said, is that plenty of data is likely already in place for them to begin the AI journey, and that it probably doesn’t need to be in particularly great shape, either.
Cohen said that because distributors inherently need two pieces of data for their business to work — who bought what, and where they are located — they are likely already in “a good place to start.”
Noble added that MRO distribution is “transactional” in nature — and therefore “nearly perfect for AI applications.”
“Chasing perfect data is really difficult,” Noble said. “There’s a place to get started now.”
Witcpalek said that so much of AI is simply “cleaning up your own data,” that the availability of that data — rather than good numbers or advanced technological capacity — is the key to beginning the AI journey. He said rather than perfecting their data, distributors should ensure that their data is entered into a single “lake.”
Once things become digital, Witcpalek said, “incredible” things can happen.
Not only do vendors not expect company data to be perfected, but the capability for AI systems to handle imperfect data are constantly getting better and better: adding additional flexibility to handle a complex array of existing systems.
Dealing with a Changing Environment
If AI seems intimidating for company executives, it has the potential to be downright terrifying for their employees, who are facing down the prospect of a dramatically different workplace — and even questions about the security of their jobs altogether.
Cohen said that dealing with “change management” is a far larger hurdle to embracing AI than any issue with data; he estimated that some 80% of the work “is actually getting your salespeople to do something different.”
He stressed that company leaders need to be clear about the goals of implementing AI, along with the metrics that employees will be held accountable to in a new environment. Proton, he said, doesn’t see AI — or anything else, for that matter — killing off the outside sales rep anytime soon, while inside sales personnel should be able to have AI take on tasks that would allow them to spend their time actually talking to customers.
Others said that AI should be thought of — by executives and employees alike — as a tool, rather than a danger.
Davoll said that technology vendors are investing in “explain-ability” — metrics that provide answers using tangible numbers, rather than just additional explanations from people.
Those answers, Noble added, should address questions about cybersecurity, while Cohen said that AI models effectively “hallucinating” information has not been a problem. When issues do arise, he said that people tend to understand that AI “doesn’t get it right every single time” and that it is not, as Noble said, an “out-of-the-box magic wand.”
“It does take time,” Noble said. “You’re going to have some failures.”
Witcpalek said that company decision-makers need to be honest about their efforts to get more efficient using AI, while addressing questions about job security and the future of the business head-on.
“[Employees’] value is doing things that machines cannot do,” Witcpalek said, particularly forging relationships with other people — customers, in particular.
“It’s likely going to replace a lot of what you’re doing,” Witcpalek said. “And that’s okay.”
Noble said AI would protect the scalability of distributors and of the distribution in general; as the next generation of workers moves into the sector, AI will “help them do their job.”
Witcpalek, meanwhile, said that although rising automation and efficiency has historically resulted in gradual attrition, those who remain in distribution in the AI era would be among the best in the industry — and be more well-compensated as a result.
First Steps — or Next Steps
Before articulating a vision for AI to the company as a whole, or before even reaching out to potential technology partners in the first place, the panel said that distribution executives need to be able to articulate it to themselves.
Andy Szal/Industrial Distribution
Noble suggested getting a handle on “the outcomes that you want to have, and work backwards from there.” A partner or internal committee, for example, can help navigate a five-year path toward realizing the benefits of AI.
“Don’t wait for perfect,” Noble said. “You’re ready, and you’d be surprised that your data is ready, as well.”
Davoll, meanwhile, said that company leaders should take it upon themselves to become familiar with new technology — and, in turn, make it less daunting for their businesses. With about 100 hours of experimentation with AI, he told panel attendees, “you’ll be an expert.”
“It's not out of reach for anyone in this room.”
