The Rise of “Cognitive” Supply Chain Control Towers

Supply chain visibility platforms from companies like FourKites and project44 have become a staple in the industry and have seen great success with multi-national companies seeking to consolidate tracking data and provide real-time insights into product location, status, and workflows. Control towers offer an extra layer to this, giving users the ability to action resolutions and collaborate with stakeholders through the platform, without the need for external methods, such as emails or calls. Blue Yonder, Hitachi Vantara, IBM, Infor, Kinaxis, Microsoft, o9 Solutions, Palantir, SAP, and Siemens, among others, have all developed their version of a control tower, offering a way to consolidate and act on the mass of supply chain data available to organizations.

At the end of 2022, ABI Insight “Microsoft’s Supply Chain Platform Goes Live, Joining a Competitive Market of Solutions Seeking to Tackle Supply Chain Fragmentation” explored Microsoft’s entry into the control tower market. Since then, control towers have been expanding in terms of their capabilities, driven largely by the wave of developments in Artificial Intelligence (AI). With increasingly sophisticated use of AI within these solutions, the term “cognitive” control towers has emerged, with indication of a shift from control towers acting as decision support tools to decision makers.

Some of the world’s largest organizations and government agencies have continued to tap into control tower solutions as a way to consolidate data, develop a single source of truth for their supply chains, and create a platform for stakeholder collaboration. Oil company Castrol is leveraging Palantir’s Foundry platform for digital twin capabilities and actionable insights to help combat supply shocks; PepsiCo is aligning its global stakeholders through the o9 Control Tower; and logistics provider DSV has partnered with Kinaxis to deploy its Command & Control Center as a tool to unify internal and external divisions as it continues global expansion.

The common driver for deploying a control tower lies in the need to consolidate information, unify stakeholders, and leverage multiple signals to generate a strategy that benefits the entire supply chain, not just a single area. Only with a network-wide view can organizations identify a resolution to a disruption that brings the whole network back to the status quo as quickly as possible. With such vast swathes of data, AI has been integral to processing and organizing these data, carving out actionable insights that would be difficult for a human to spot. But as vendors continue to work AI into control towers, more functionalities are shining through:

• Network Optimization: By analyzing an organization’s full supply network, AI can identify the most optimal way to organize inbound and outbound logistics. The ability to analyze vast sets of data and run what-if scenarios unlocks new potential for efficiencies, savings, and Carbon Dioxide (CO2) reduction.
• Simulations and Scenario Analysis: With both a digital twin of the supply chain and ability to analyze greater amounts of data inputs, “cognitive” control towers can run more advanced simulations to identify the right strategy to tackle an issue. AI integration enables solutions to learn continuously from experience, while factoring in changing events.
• Decision-Making Capabilities: Control towers typically offer decision support through the insights provided, with human involvement still required to action the resolution. “Cognitive” control towers can provide automated decision-making, greatly speeding up the ability to resolve disruptions.

Taking a network-wide approach with the data processing support of AI is creating significant opportunities. From a pro-active standpoint, an organization looking at its full network, rather than individual suppliers allows it to identify much larger savings through methods like groupage, stock consolidation, backhaul, fronthaul, and travel reduction. For example, Siemens has incorporated a planning and optimization function into the Siemens Digital Logistics platform, allowing users to see “as is” and “optimized” arrangements. Supplier locations, consolidation hubs, and subsequent travel distances are mapped and then rearranged by the system, producing identified savings for review. Supply chain planners can then collaborate through Siemens AX4 platform with relevant stakeholders to change their logistics arrangements and realize the savings.

As with any AI-enabled application, the results are only as good as the data provided to the system. Modular control towers that are implemented to optimize one area of the supply chain, such as an inventory or fulfillment control tower, are limited by the lack of data signals coming from adjacent functions. In the same way, an organization’s end-to-end control tower can remain limited if it does not tap into a cloud network that obtains data from other control towers deployed by other companies and industries. While a pharmaceutical supply chain is different from an automotive supply chain, for example, both will be affected by the same bottlenecks, geopolitical disruptions, and trade restrictions. Leveraging extensive cloud networks feeds a control tower with a more comprehensive stream of data, delivering greater intelligence to either the decision support or decision-making function.

There is also a question as to whether people are comfortable allowing systems to evolve from decision supporters to decision makers. While AI-enabled control towers can take more signals into account and produce decisions quicker, potentially avoiding disruption that may occur in the time it takes for stakeholders to decide on a resolution, relieving such control is not something companies are always willing to do. Implementing AI control and guardrails is something that Palantir has implemented in its control tower, allowing users to define what actions the Large Language Models (LLMs) and AI can take and where to include a human-in-the-loop. Such restrictions can help ease AI adoption and ensure compliance with applicable regulations.

In the same vein, users only want to pay for what they use. Control towers are intended to be omnipotent, but different users will require much different levels of depth and granularity. Siemens has opted for a pay-as-you-use model in its solution, only charging customers when they decide to expand the functionality of the control tower. Scalable Software-as-a-Service (SaaS) that fits and changes to meet individual needs will continue to be imperative to wider adoption.

While the reality remains that control towers are only as good as the data provided to them, end users shouldn’t wait for data streams to be perfect before they consider deploying control towers. With augmentation from AI, solutions are becoming better at dealing with imperfect data and can even help correct issues over time, helping companies to start accessing the potential in their data and start realizing the benefits that control towers offer.