How does artificial intelligence (AI) enable digital transformation in distribution? And what should distributors keep in mind when evaluating and implementing AI as part of their overall strategy?
Distribution Strategy Group’s Jonathan Bein recently tackled these questions in a recent Technology Leader Panel.
The panelists included:
- Alex Witcpalek, CEO, Continuum
- Jared Helenic, Product Specialist, Infor
Missed the webinar? Watch it on-demand.
The Value of AI Today
Science builds upon established knowledge. That’s what got us AI; the cloud infrastructure and big data helped enable AI into software as a service (SaaS) models that anyone can use. This democratization gives distributors access to tools today that can automate mundane tasks.
Witcpalek said the value of current AI models is in strategically improving existing processes. “That needs to be step one, with AI encapsulated inside of your tools, technologies and the systems that your team interacts with.
“Once you do that, you can apply AI to automate decision-making, setting you up for top-down innovation. Right now, we’re just in the building blocks in the beginning of how AI can help leverage data and insights for the digital transformation.”
Helenic said AI can help distributors “make decisions based on historical data and seasonality data entry points.” Customer engagement improves when the AI takes past purchasing patterns and uses them for the cross-sell or upsell. AI improves how quickly teams generate proposals and respond to customer needs.
“I think the other area of impact is on the individual seller, making them as efficient as possible. How quickly can they have the right product recommendations when they’re giving an order?” he said. “The right part is sometimes determined based on where you’re selling. North Dakota may have a different valve than South Carolina and so on. A less tenured rep may not know that off the jump.”
Where to Start with AI
So, where should distributors begin to understand where to apply AI in their own businesses?
“The most successful people I’ve seen with their digital transformations go through a readiness assessment,” Helenic said. “That’s where they go through each business unit, marketing, finance, operations, etc., and they mark down what’s taking most of their time; what are these processes?”
Mapping out these workflows and then using AI to focus on the biggest pain points are what will improve a distributor’s productivity.
Witcpalek said that once a distributor identifies cumbersome workflows, the next step is to identify and implement tools leveraging AI to support continuous improvement.
The Infrastructure to Support Your Transformation with AI
Even the oldest legacy ERP platforms can connect with third-party vendors to apply AI in a business. “You could put an API (application programming interface) wrapper around that system, which now gives third-party technologies access. There are strategies so that a company that may still run on old technology could very quickly gain access to or the ability to leverage AI when they maybe didn’t think they were ready,” Helenic said.
Increasingly, software providers (core ERP and third-party vendors) are integrating AI into their tools. This opens the door to even small distribution companies getting access to the power of AI. Helenic said as a result there’s a lot of opportunity to execute and leverage AI solutions if there are boundaries on your current tech stack.
He also pointed out an important part of building an infrastructure to support AI is to invest in internal teams to ensure they’re up to speed with the best use cases for these technologies.
Witcpalek agreed. “Outside of the technology, the investment on the people side is important. The new role is something called a data scientist. This individual understands Python mathematics, but we’re missing business knowledge. You can hire someone right out of college. Still, it will take time for them to work with your business folks to understand the right questions, to take that knowledge and translate it into a mathematical equation in a formula, and then use that to make it embedded in those machine learning decisions.”
The cloud remains perhaps the most critical component of building an AI-supportive infrastructure. “If you’re buying on-prem servers,” Witcpalek said, “it’s going to be difficult to manage because the amount of compute resources you’ll need will increase exponentially.”
For distributors tempted to build a custom infrastructure to support AI, both panelists agreed that the costs of developing large language models and the computing power to power them are out of reach for the typical distributor.
“A big push for why the cloud is going to be a little bit easier is you’re not going to have the amortization as you do with hardware,” Helenic said.
Witcpalek agreed. “Portions of AI are democratized in the sense that you can take subsets of open-source AI models and build on top of them so that you do not own all the compute power internally in your organization. You’re just building on top of that and enhancing it. I think that’s probably the more realistic strategy.”
Is Your Data AI-Ready? The Answer: Yes
Data is the fuel for AI. But don’t let the state of your data deter you from starting your AI journey. Most distributors use historical data that live in separate buckets for everyday tasks and forecasting. Helenic said the key to mastering these silos is to create a data lake, or a centralized holding repository for all information in the business.
“The big first step is knowing what is centralizing and then working with your data science team or a vendor to evaluate what’s missing. A lot of times, it’s not going to be as much as you think, and it’s going to be easier to pull in other things like product attributes, which will make it a lot better.”
Witcpalek added this caveat: “You might struggle to find someone who gets really excited about being able to try to manage all that. Then they’re trying to build AI on top of wrangling data. Keep in mind, that’s just getting data in a single place. It’s not saying your data needs to be clean. Your data doesn’t need to be clean to get started with AI.”
Helenic encouraged distributors to not be afraid to try and fail. “That’s a big part of this. I’d much rather see people try and then work with a good partner to make sure that they can be successful.”
Missed the webinar? Watch on-demand.
