Arkestro brings AI to ERP
We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 — 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
The supply chain — the once forgotten part of the economy that was largely taken for granted until the pandemic — is being reimagined with predictive machine learning algorithms. Arkestro, the San Francisco-based makers of what it calls a “predictive procurement orchestration platform”, announced a successful series A funding round today that will raise $28 million to bolster their efforts and expand their product.
The process of managing the often complex steps that turn parts and raw materials into finished goods has been a job for digital computers for decades. Companies like IBM, SAP and Oracle built empires by tracking goods, invoices and bills of lading, a process they often described with the acronym of “enterprise resource planning” or ERP. They were often so successful that everyone forgot they existed. Then the COVID-19 pandemic sent shockwaves through industries as managers up and down the supply chain worked to handle holdups, glitches and delays.
Arkestro is bringing a mixture of modern artificial intelligence algorithms to the world of ERP. It wants to use a combination of machine learning, game theory and behavioral science to make the supply chain more efficient and better able to withstand failures.
“If you talk with most procurement and supply chain leaders in the global economy today, they’ll tell you that the number one challenge is there are just not enough qualified people to do the work of collecting information required to support supplying goods and services at competitive prices for large enterprises,” said Edward Zagorin, Arkestro’s CEO.
A large part of Arkestro’s solution focuses on allowing algorithms to help people with pricing and price discovery, a challenge that is also only growing more complex now that inflation is putting more pressure on the enterprise to charge the right amount. When prices are stable, suppliers can work together for years without renegotiating prices. But the specter of currency erosion makes it crucial for companies to find the appropriate price.
Arkestro offers a predictive pricing model that tracks the relationship between both sides of the transaction while also folding in data from others in similar marketplaces. It also tracks anomalies and disruptions that can send pricing shocks.
“We don’t just rely on what the previous price purchase was,” said Zagorin. “We actually integrate a number of factors.”
The algorithm may consider details like the locations of both sides of the transaction, the frequency of transactions and the prices paid by other similar companies in similar contracts.
The main system also includes a recommendation engine which is similar, at least in spirit, to the customer-facing recommendation systems that are built into ecommerce sites. Their approach, though, is adjusted to work better with the smaller dynamics in industrial supply chains were there are many fewer sellers and buyers.
The team at Arkestro is heavily influenced by much of the economic theory on finding the best pricing solution for both sides. In an interview before the announcement, Zagorin mentioned classic game theory models like the prisoner’s dilemma and illustrated how Arkestro wants to help both sides of any transaction find the best long-term solution.
When Arkestro enters the negotiations as a third party, the dynamics can change. Sellers and buyers can think beyond the zero-sum nature of price negotiation when the algorithm in the middle of the transaction is suggesting a price.
“When it’s the machine making the offer instead of the supplier, it actually flips the game theoretic logic of the motivation to accept,” Zagorin said. “Suppliers are more motivated to accept offers that are generated by Arkestro because they know that other suppliers have the ability to be shown the same offer, and so they don’t want to miss out to a competitor.”
How AI research folds into ERP
Many in the field are also bringing artificial intelligence to business decisions, but maybe without the same focus as Arkestro. Oracle, Microsoft, IBM and SAP all have strong programs in artificial intelligence (AI) research and they’ve been folding it into their ERP offerings for a while.
Microsoft’s Dynamics 365, for example, is one of the platforms built by Microsoft for managing business processes. The Azure platform offers both the opportunity to work directly with predictive analytics directly or by activating the versions that are pre-integrated with the Dynamics 365 system.
Their systems, though, are more general and could be applied to all types of business statistics, not just supply chain decisions. They’re more open tools, and Arkestro is focusing on building the optimal algorithms for one important part of business that’s very much in the spotlight today.
“If you look at the past 30 years of procurement, a lot of procurement teams were primarily administrative and they were graded solely on cost saving,” said Zagorin. “If you look at where procurement and supply chain are today, they are in many cases on the front lines of every incoming issue from the C-suite, whether that is driving towards sustainability and ESG impact or dealing with sanctions.”
The funding round was led by New Enterprise Associates (NEA) , Construct, Koch Disruptive Technologies (KDT) and Four More Capital.
In the announcement, Jeff Immelt, a venture partner at NEA praised the company’s “differentiated, AI-powered solution” for simplifying the supply chain management.
“We see significant potential for Arkestro and its technology platform in the supply chain; notably in terms of the value it will bring to the procurement market to help enterprises realize increased efficiencies and savings in their operations,” said Byron Knight, COO and managing director at Koch Disruptive Technologies. “KDT and Koch Industries have numerous investments and operations across the supply chain, and we’re excited to help Arkestro explore the potential applicability of its solutions.”