How AI is transforming inventory management

About two years ago, ThredUp started using visual tools that it developed in-house which can pinpoint an item’s brand, size, washing instructions and other information in a matter of seconds. After the system was implemented about two years ago, it resulted in a 10% productivity lift right away.

“It makes it so the team member on-site doesn’t have to work their way through a really large hierarchy of tagging data,” Homer said. “They’re just doing the work to handle the garments and take these pictures and make sure it gets the quality inspected. And the system behind the scenes is doing all the automation.”

In the supply chain world, artificial intelligence is helping some companies streamline how they manage inventory. McKinsey in a March 2025 survey release found that about 78% of all organizations were using some form of AI in their business in July 2024, up from 55% in 2023. But out of consumer and retail goods businesses surveyed, just about 14% of companies said they were regularly using generative AI for supply chain and inventory management.

Some who are, though, are saying generative AI or AI-powered algorithms are helping them save money on extra shipping fees or expedited orders. Companies like Bansk Beauty and diaper brand Kudos are using AI to zone in on inventory needs and overall counts, helping them ensure they’re making just enough to keep up with demand. Beyond getting more accurate figures in place, the companies are finding workers are freed up for other tasks or to simply meet more demand.

For ThredUp’s part, investing in AI infrastructure is a key part of how the company intends to grow as a secondhand clothing marketplace. Already, the AI-driven tagging system helps get more items up for sale faster. During the company’s most recent earnings call earlier this month, CEO James Reinhart said the company has 9% more fresh items listed quarter-to-date than the year prior. “With ongoing innovations in processing technology and AI, we are extending our advantage even further while we grow capacity and reduce costs,” Reinhart said on the earnings call.

Other behind-the-scenes listing tools, like 360-degree photos, have increased 30-day sell-through rates by up to 12%. Homer said the camera setup not only gets items up faster, but it also has the potential to generate AI-powered detail shots of buttons, fabric or other characteristics to give shoppers more details about the product they’re buying.

“In the last couple of years, we’ve been able to find that stride with the technology improving and being ready for it,” Homer told Modern Retail.

But Homer said the tools are ever-evolving and have to keep up with the company’s own changing needs. ThredUp is processing used clothing that could come from any number of brands, and so its tools are consistently getting new inputs to recognize those tags and logos — and recognize them if the branding changes. Homer said ThredUp mitigates some of this by having a reviewer look at what the AI tools are pulling and correct any inconsistencies. There’s also a way to go with AI being able to detect what kind of fabric something is made of.

But overall, the company is holding millions of data points that, when combined with algorithms, can help improve its item processing. As of early 2025, ThredUp is also beginning to roll out item measurements that are captured based on an image alone.

“The accuracy is an ever-present challenge, but it’s more precise and accurate coming through the systems that we built now than it was in the past,” he said. “There’s this combination of our team members and the systems working together to sometimes automate, but then sometimes assist, and that helps train the system to get better and better over time.”

For brands that manufacture their products, AI inventory management tools can help zero in on how much a company needs to make, and where its products need to go, without any excess inventory collecting dust in a warehouse. Ilan Tagger, chief of staff at the diaper brand Kudos, said AI-driven tools can crunch more data faster — and with more accuracy — than human counterparts. “It involves quite a bit of math and modeling. I don’t think the average planner would be able to calculate it on their own through Excel,” he said.

Kudos uses a vendor-built AI tool called EverX to help its inventory management and forecasts. It’s able to help the company figure out its demand needs across each SKU and each geographic zone to determine inventory needs for warehouses in real time. This can particularly help Kudos figure out where it may be able to decrease its spending on fulfillment or where it needs to add more products when it comes time for replenishment. For example, if a particular SKU has sold out in one region, it can direct 70% to the nearest warehouse and the remaining 30% to another location.

Tagger said the system has also been useful in making sure that inventory reporting at warehouses matches what it should be, based on how many shipments in and out are on the schedule. This helps avoid costly, time-consuming miscounts.

“In a physical warehouse, when [miscounts] would happen before, we would be too deep into the miscount, and it would already be applied to our forecasting and what we were sharing outwardly,” he said. “Now, we can at least get ahead of it and catch it mostly in real-time.”

Liran Golan, chief supply chain officer at Bansk Beauty, also said forecasting has been the most helpful use case for AI in the company’s logistics operations. The company houses multiple beauty brands, including Amika hair care. In December, Bansk started using algorithms for forecasting and inventory management. These look at the confluence of past shipments, sales data, promotions and other data points to predict how much will be sold in the next period and through what channels.

As a result, the company is holding about 50% less inventory than it was a year ago, Golan said. That’s because the AI-driven forecast also helps the company understand how much of its inventory to put in which distribution centers, saving on fees to expedite more orders to certain locations. “Having a good forecast reduces my inventory requirements, which means I’m tying up less networking capital for the business,” he said.

And while logistics operations have long used forecasting to help understand inventory needs, Golan said that the AI-driven algorithms free up workers’ time for other tasks.

“People think AI is going replace supply chain and I’m not going to need people anymore, but that’s not how I see it,” Golan said. “I see AI as taking away the transactional, mundane, non-value-added work to free up time to actually do value-added work.”