How this startup helps retailers navigate supply chain chaos

The AI supply chain startup Altana was formed in early 2019 around a forecast of global chaos. As part of an early exercise, the three co-founders each traced out what they thought trade would look like in the next few years.

“We all wrote some version of, like, the world is gonna fall apart,” CEO and co-founder Evan Smith told us, specifically recalling predictions of more trade wars spurred by the economic rise of China, climate dislocation, and sovereign debt crises.

Seven years later, chaos has become the norm. Soon after Altana’s founding, the coronavirus pandemic ramped up a period of global trade turbulence that hasn’t eased much since. More recently, whack-a-mole tariffs and war in the Middle East have roiled the globe-spanning trade routes on which many retailers rely.

Altana aims to help businesses, logistics providers, and governments navigate these uncertainties. The company runs a data platform that shows the operating facilities and ownership of companies and the flow of goods between them, according to Smith.

“You actually get to see this living, breathing, bottoms-up view of raw materials coming out of the earth and then being transformed into components, and then turning into, like, an iPhone,” Smith said.

On top of that Google Maps-like display, clients ranging from L.L. Bean and Skims to shipping giant Maersk and US Customs and Border Patrol can use analytics dashboards, predictive AI, and other digital tools to manage their supply chain networks. Altana also claims to offer more visibility into a “genealogy of the goods,” from subcomponents to “sub sub subcomponents, the raw materials,” Smith said. The startup had raised $344 million at a nearly $1.3 billion valuation as of last August, according to PitchBook data.

“The big headline bet we’ve made over the last 18 months is in automating all cross-border trade functions for all three of those parties. So in the government case, you’re trying to enforce the law. In the enterprise case, you’re trying to comply with the law. And the logistics provider, we’re trying to facilitate that whole network,” Smith told us at Altana’s Brooklyn headquarters. “And we have a ton of AI workflows, agents that are doing all three of those jobs in a connected, collective network.”

Polygons of disruption

When surprise US and Israeli strikes on Iran at the end of February set off widespread war in the Middle East, Smith said Altana clients could game out how the conflict might interrupt their flows of goods through the region.

Clients could “basically draw a polygon around the Gulf,” then simulate different types of disruption and the downstream effects they might have, Smith said. He mentions questions like: How do shortages of helium, an irreplaceable need for chip fabrication, impact the semiconductor supply chain? What does a lack of the chemical compound urea mean for food and fertilizer?

“You can actually get to the expected value of a loss of a business interruption from this, that, or the other thing—the supply chain going ‘poof’,” Smith said. “So every time there’s a big disruption of the system—obviously Iran right now, but Russia and Ukraine, the tariffs—you get these big waves that slosh around the whole global network. And our system tells you what parts of your business are exposed to those things.”

A new era for supply chains

The pandemic, global conflicts, and tariff shocks have pushed retailers and shipping companies to adopt AI and other technology to better map and analyze supply chains. Furthermore, new laws around forced labor and mineral sourcing require more transparency into different levels of the supply chain.

A little over half of logistics leaders Gartner surveyed recently ranked advancements in AI as the top driver in redefining supply chain strategy in the next two years.

Altana is one of several companies attempting to meet this new demand with AI-powered supply chain tools. One of the biggest tasks is simply making sense of a huge variety of transaction data from different sources with varying formats, languages, and human errors, according to Smith. That’s done with a proprietary combination of machine learning models and, more recently, LLMs, though they aren’t “the right tool for the job for a lot of parts of that AI pipeline.”

“The challenge is you have these hundreds of millions of companies, you have these billions of transactions between them, hundreds of places on earth, and there’s no unique identifiers,” Smith said. “It’s not like an email address or cell phone [number]…So for the machine, you’ve got to read Chinese and Spanish and English, and read an address and recognize that somebody accidentally put the company name inside of the address…and then there’s a gazillion ways to describe physical goods. So how do you make sense of all that stuff and then turn it into one canonical, unified representation?”

While trade was previously seen as a predictable, back-office function for many retailers, Smith said it’s becoming more instrumental in how retailers think about overall strategy as tariff, compliance, and geopolitical situations evolve.

“Some of our retailer, brand, [and] manufacturing customers are starting to see trade as a top-of-mind, board-level, strategic issue,” Smith said. “They’re asking, ‘How do I bake this into the product design and engineering and sourcing steps?’”