Northwest Meat Co. opened its doors in 1956. The business was originally located in Chicago’s Six Corners area. In 1963 the processor/distributor moved operations to the Fulton Market area and in 2014 moved to its current location just west of the Fulton Market area.

The company employs 21 people and operates out of 15,000 square feet including 12,000 square feet of processing including cooler storage and dry age storage, along with 3,000 square feet of office space.

“We are a processor,” said Andrew Neva, vice president of operations, and the third generation of the family-owned business. “We process beef, veal and lamb, and we grind our own ground beef in house, and we make our own homemade Italian sausage.”

Northwest Meat does not slaughter or break carcasses. It will bring the meat in and sell it as a primal or portion it into steaks, chops, ground beef, sausage etc., to foodservice exclusively. Customers consist predominantly of restaurants, hotels and country clubs throughout the Chicagoland area and a small portion of business in southern Wisconsin.

“We do some stuff out of state, it’s going to be a drop shipment through UPS, usually, that’s a pretty specialized product that somebody can’t get,” Neva said.

Northwest Meat buys its beef predominantly from Nebraska Beef in Omaha and also distributes Piedmontese beef to Chicago and offers customers a dry aged beef program. Pork comes from Indiana Kitchen and Seaboard Farms. Veal is purchased from a Dutch producer, and lamb comes from either Colorado or Western Pennsylvania.

As a successful, small, family-owned business entering into an ever-changing meat processing industry, Northwest Meat Company realized it needed something to maintain its position with its customer base and within the industry without depleting its margins.

Northwest Meat Co. buys meat almost exclusively from the Midwest for its foodservice products. (Source: Choco)

Incorporating Automation and AI

Until the recent past, Northwest Meat Co. filled its orders the traditional way. Customers called during business hours, gave their order to a person on the other end who then manually entered the order into the company’s ERP (enterprise resource planning) system. The order printed out and was filled.

Northwest Meat also took orders via email and an after-hours voicemail. Those orders were processed in the same manner the following business day. That procedure takes two precious resources, time and labor. Neva and the company decided to partner with Berlin-based Choco, which has an office in Chicago. Choco is an online platform that enables restaurants and their suppliers to mutually streamline their order processes.

“If we are going to remain relevant, and if we want to continue to do things, and we’re going to continue to feed the world, because if we get behind the curve of technology, it’s going to be bad,” Neva said. “We had to act. So that’s why I really partnered with this because it was essential to my operations.”

The time savings on order processing Choco provides are significant.

“It takes a long time, and it gets harder and harder to find people willing to process them,” Neva said. “It’s going to take an hour and a half, two hours. You’d be surprised how fast it is. I did it the other night. I was done in 20 minutes.”

Neva discovered Choco through one of Northwest Meat’s sales reps who was pitched while working part time at a restaurant. She told Neva about it, and Neva was impressed.

“So, I started sitting down and talking with them, and it turns out they literally just set up shop in Chicago,” he said. “We might have been one of their first-ever contacts on the wholesale side.”

During the construction and integration of Choco and Northwest Meat’s systems, Northwest Meats and Neva, provided Choco the opportunity to create a template for American wholesalers and distributors because the European models Choco dealt with differ from those in the United States.

“It kind of was like an organic kind of thing, like we just kind of crossed paths, and we just kind of hit it off,” Neva said. “And now it’s kind of led to, you know, they’re doing this AI automation, they’re doing payments, they’re doing many different things.”

The integration between Choco’s system and Northwest Meat’s ERP system took some time and was tedious. Choco had to pull all the item codes, and Northwest Meat had to make sure they were correct, and software to allow the two systems to communicate had to be developed.

How it works

There are four different ways Northwest Meat Company takes an order, and Choco handles three of them. There’s the original way, a customer calls during business hours and an employee writes down and enters the order. The second is through email, the third is voicemail and the last order entry method is through the Choco app.

“Certain customers use the Choco app to place their order,” Neva said. “There’s a custom Northwest Meat order guide on their account. Once that’s in, it gets shot over to our ERP without the need for approval.”

Email and voicemail orders are where the AI come into play. The Choco system will interpret email orders and voicemail orders, but transcriptions can be tricky depending on many variables such as individual speech patterns on voicemail, and spelling and grammar in emails, just to name a few. These orders are placed in a staging queue and must be approved before going to the ERP to be printed out as orders ready to fill.

“Once it’s approved, every 10 minutes it gets cycled into our ERP system to be printed out,” Neva said. “So, you don’t even have to send it. You’re just in there, approve, approve, approve. If nothing is ready to go it doesn’t do anything.”

Northwest Meat Co. employees can approve orders in the system at any time. Every 10 minutes the system will check for approved orders, if there are any it will send them through to the ERP, but only the approved orders. If there are no approved orders, it will check again for approvals in 10 minutes. The cycle is continuous as the system uses AI to learn through repetition and history. As it transcribes the email and voicemail orders and sees corrections and approvals, it will begin to make the corrections on its own allowing faster approvals over time.

“You might have to address mistakes between 10 to 50 times, but eventually it will get it, and then it’ll just correct it on its own,” Neva said. “It should never have a problem once you’ve trained it properly.”

Andrew Neva, vice president of operations, received a bachelor's degree in meat science from the University of Illinois. (Source: Choco)

How it’s working

Northwest Meat Co. has been working with Choco and using the system for eight months and because of the complexity of integrating multiple systems, there have been challenges along the way.

“When you have three or four different facets trying to talk to each other, and you’re trying to centralize them all into one location, that makes it difficult, things happen,” Neva said. “However, if you look at it from 30,000 feet, it has been amazing, it has been such an asset.”

Neva said the AI part of the system is still learning, but he has been pleasantly surprised on multiple occasions by the ability of the system to pick things up quickly, and also be patient with the system when it has made some corrections that don’t make sense.

“The other day I was looking at orders and thought it’s going to get this thrown off,” Neva said. “I went and checked it, and it actually got it. There’s also been times where I’ve thought, ‘what is this thing doing?’ but that’s how it goes and that’s to be expected. When you look at the majority of the time, it’s getting it 80 to 90% of the time. There’s always going to be that 10 to 20%.”

The system continuously learns. If a new customer places an order, the AI will have no history or data on that customer. It will transcribe the order and place it in the queue to the best of its ability, but perfection cannot be expected, especially under those circumstances.

Neva also noted that a good portion of the responsibility falls on the user. The AI can only learn what the user teaches it.

“Honestly, it never stops learning,” Neva said. “Every time you keep hitting confirm, confirm, confirm, it’s just continuing to gather and gather more data.”