All of the “Is this a bubble?” talking about artificial intelligence, supply chains and logistics industries has become a breeding ground for the apparently authentic use of technology. Flexport, Uber Freight, and dozens of startups have developed a variety of applications and attracted Blue-Chip customers.
However, while AI helps the Fortune 500s fill revenues (and justifies the next layoff to Wall Street), proper use of the technology has proven useful for small businesses.
Netstock, an inventory management software company founded in 2009, is committed to just that. Recently, we have deployed a generic AI-powered tool called “opportunity engines” that slots into existing customer dashboards. The tool extracts information from your customer’s enterprise resource planning software and uses that information to make regular, real-time recommendations.
Netstock claims the tool saves thousands of people on these businesses. On Thursday, the company announced that it has provided 1 million recommendations so far, with 75% of its customers receiving opportunities engine proposals worth more than $50,000.
With an appetite in mind, one of those customers (a family-owned, 65-year-old restaurant supplier) initially felt uneasy about the use of artificial intelligence products.
“Old family companies don’t trust blind changes very much,” Chief Innovation Director Jacob Moody told TechCrunch. “I would have gone into the warehouse and said, ‘Hey, this black box will start managing it.’ ”
Instead, Moody has pitched netstock AI internally as a tool that allows warehouse managers to “choose to use or not.”
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
Moody says it helps to avoid mistakes as he sifts through countless reports his staff uses to make inventory decisions. He admitted that the AI summary of this information is not 100% accurate, but said it “helps create signals from noise” especially outside business hours.
A “more serious” change that Moody noticed is that software, which has become part of Bargreen Ellingson’s lesser warehouse staff, has become “more effective.”
He highlighted the employees in one of Bargreen’s 25 warehouses, who have worked there for two years. The employee has a high school diploma but no university degrees. He said the employee will train him to understand all about inventory management tools and the forecast information Bargreen uses to plan inventory levels will take time.
“But he knows our customers and knows what he rides on the truck every day, so for him, he can look at the system and get this mediocre AI-driven insight and get a very quick idea of whether it makes sense or doesn’t make sense,” he said. “That’s why he feels empowered.”
Barry Kukkuk, co-founder of Netstock, told TechCrunch that he understands his hesitations with new technologies, especially since many of his products are mediocre chatbots that are essentially attached to existing software.
He attributes the early success of Netstock’s opportunity engine to some. The company has over 10 years of data from working with retailers, distributors and lightweight manufacturers. That data is strictly protected to adhere to the ISO framework, but it drives the model that makes recommendations. (He said Netstock uses a combination of open source communities and private companies’ AI Tech.)
Each recommendation can be evaluated by giving a thumbs up or giving a thumbs up, but the model is enhanced by whether the customer takes the proposed action.
This kind of reinforcement learning can lead to strange and sometimes harmful outcomes when applied to something like social media, but Kukkuk said he is chasing a variety of incentives.
“I really don’t care about my eyes, do you?” he said. “Facebook and Instagram care about their eyes, so they want you to look at theirs. We’re saying, “What are the customer’s outcomes?”
Kukkuk is wary of expanding these interactions due to current limitations of generator AI technology. It may make sense for customers to have a conversation about whether NetStock’s AI and recommendations will be useful, but Kukkuk said it could ultimately lead to a breakdown of accuracy.
“Walking is a tightrope walk because the more freedom you give to the user, the more freedom you have to give them a larger language model to begin hallucinating,” he said.
This explains how the deployment of the engine is an opportunity to be made into a typical customer dashboard on NetStock. The proposal is prominent, but is easily rejected. Google Docs has 20 AI with the user’s throat, but it’s not.
Moody said he was grateful that AI wasn’t on your face.
“We didn’t let the AI engine make a decision on stocks that humans didn’t see, and they said, ‘Yes, I agree with that,'” he said. “If you reach a point where you agree to 90% of what you’re proposing, you might take the next step and say, ‘Give control now.’ But we’re not there yet. ”
This is a promising start when many enterprise deployments of generative AI seem to go nowhere.
But when the technology gets better, Moody nevertheless said he is worried about what it means.
“Personally, I’m scared of what this means. I think there will be a lot of changes, and none of us are really sure what it will look like in bargreen,” he said. This could lead to fewer data science experts on staff, he suggested. But even if that means moving those employees from warehouses to corporate offices, he said it’s important to preserve knowledge.
Bargreen needs people who “have a deeper understanding of theory and philosophy, and be able to streamline how and why Netstock is making certain recommendations,” and they need the wrong path “to avoid going blindly.”
Source link
