Wild Bull Services seeks AI breakthrough

Too often, companies simply add AI without a clear goal—in many cases doing so because their board has directed them to do so. But most experts would agree that is the wrong approach. You wouldn’t add an electric truck with a 200-mile range to a 600-mile route, so why are you adding AI in an area where it doesn’t make sense?

There are thousands of companies now pitching their AI prowess to businesses. If you have a pain point, there is a technology supplier that says it can solve it with artificial intelligence.

One of the areas ripe for AI is the payables. Wild Bull Services, a top-three importer of beef to the U.S., went down the AI rabbit hole with AI-payables provider Stampli.

And rabbit hole maybe the appropriate term as Wild Bull started with payables, but is now exploring how AI can transform its import documentation.

Getting started

Brian Anthony, president/owner along with his sisters Rochelle and Danielle, of Wild Bull Services, said his firms AI journey is exciting, but it all began with the company’s payables. Wild Bull imports beef, butter, pork and shrimp, so even its payables process was in need of automation. The work of processing payables was time-consuming, Anthony said, and since the documents were standardized, ripe for automation.

“We didn’t lose any people; we redeployed,” he tells Supply Chain Management Review. “The intent was not to get rid of people, it was to redeploy resources better.”

Anthony says 80% of Wild Bull’s payables are now automated through Stampli’s Billy Bot AI. He doesn’t believe it will get any higher, and that is an important point for companies to consider. AI is not a 100% solution, he says. The remaining 20% of payables are handled by humans, in some cases because there is an advanced approval process required, and in others because it is the nature of how Wild Bull prefers to verify. Anthony says manual checking is required to ensure the system is operating correctly, and when it produces AI hallucinations, there is a correction component so the system doesn’t make the same mistakes in the future.

“We’re not finding errors like we did out of the gate,” he says. “But you are not teaching it, you are just correcting it so it learns from its mistake. It’s not coding … you’re just correcting a mistake so that it learns from that correction.”

Future projects

Payables, though, is not what has Anthony excited about Wild Bull’s AI use. It is a project that has yet to take hold but could transform the company’s time commitment.

“A bill of lading number is a bill of lading number. A container number, every single container has a specific number on it, it’s not duplicated. Those things are standardized, but what’s not standardized is where those [numbers are located and what they are called],” he says.

Wild Bull, along with Stampli, is working to develop an AI-based platform for automating the ingestion of data from import documentation. Anthony says this is not a simple process since one document may list a bill of lading number and another may say BOL#. Since there is no standardization, training an AI to accurately ingest the data correctly is not an easy task. Anthony looked at solutions on the market, but couldn’t find one that met Wild Bull’s needs, so he enlisted the help of Stampli, which had already successfully proven its AI tech with Wild Bull’s payables.

Inquisitive minds

“It didn’t take long until their senior developers wanted Zoom calls with me and were asking ‘what are you thinking,’” Anthony says. Those conversations are ongoing, and he says this project is a long-term goal and it could be a year or more before a solution is ready, but he is hopeful.

“Long-term, can Billy Bot can assist us in import documents, grabbing the information?” he asks. “We have 50 or 60 vendors around the world and … every single one of them does documents a little different, and in our office, there is a lot of human capital whose sole job is [grab this information] and get it into our system.”

Anthony estimates that 80% of Wild Bull’s labor is responsible for collecting and importing import information. “The data is there,” he adds, saying that working with Stampli has been great as they ask a lot of questions about the process and information needed.

The challenge with automating import documents is they need to be correct, Anthony says. Since Wild Bull deals with Customs and Border Patrol and government agencies, incorrect information can slow the process at best, and create unnecessary scrutiny.

Anthony says the biggest learnings on AI is that you are not teaching the system, but rather correcting the system. “What we need is not for us to program or code how to do something, we want a tool that is going to learn as we go,” he says. “The most expensive part is going to be the maintenance. AI to me is the learning piece. Once you get past those hallucinations on what is not correct, then you can start driving value in the business.”

Wild Bull saw that firsthand when one of its trucking company partners switched systems. Two previous companies used the same system and initially, Billy Bot assigned invoices to the wrong company. But, once Wild Bull’s staff corrected the initial mistake, it hasn’t happened since, and when the third company came onboard, the system smoothly converted to the proper invoicing.

The learning is an important part of the process, Anthony says, as the system doesn’t know what it doesn’t know. But, once it does, it is able to apply the correct knowledge in future situations. If the import documentation system comes to fruition, it should work the same way. And that is a time-saving for a company that is importing nearly 200 million pounds of beef per month.

SC
MR