AI solutions are only as reliable as the data feeding them. Before automating decisions, organizations need confidence that operational data is accurate, governed and consistently interpreted. 

In practice, that means establishing clear ownership of operational data, consistent definitions across systems and controls over how information is accessed and shared. 

“It’s one thing to run an AI agent pilot. If you turn it on and it gets a wrong signal, and you spend $100 million on the wrong thing, who’s accountable? I think organizations underestimate this,” says Massie.

As Lenovo expanded its AI initiatives, the company discovered automation increased the importance of human accountability. Teams became responsible for validating and governing operational data before AI systems acted on it. Automation followed governance, not the other way around.

Geopolitics also factors into data collection and interpretation. Data flow should be consistently reviewed and monitored, particularly when suppliers and manufacturing sites change locations. Every change in manufacturing location, supplier network or regulatory environment affects how data is collected, shared and governed. This creates new inconsistencies that AI may unknowingly amplify.  

“One thing I think people underestimate is that, [with] all the geopolitical change over the last seven or eight years, you’ve moved nodes all over the place — whether suppliers had to change locations or you’ve relocated your own manufacturing sites," Massie says. "As you do that, the whole data picture you're collecting, the way data flows, how it’s organized and constructed, completely changes as you’ve changed all the nodes in your network. It’s really hard to keep up with that.”

While AI governance helps to track and manage data across locations and suppliers, human accountability remains essential. 

Build resilience before automation

Digital supply chains frequently struggle with tariff changes, supplier disruptions, shifting manufacturing footprints and governance inconsistencies. Resilience must come before automation, and part of that is visibility.

An IAPP (International Association of Privacy Professionals) analysis on the hidden fragility of AI supply chains found that a common blind spot emerges in many well-meaning AI governance frameworks: managing risks in the vendor and third-party ecosystem. Massie agrees.

He says when companies can peer deeper into the data, they can better determine what’s been affected in a disruption — and remedy the situation quickly. 

“I think one of the big things Lenovo’s done really well is map data down to second- and third- level suppliers,” says Massie. “And having that data visible to those suppliers regularly has been a key differentiator.” 

This visibility allows Lenovo to identify vulnerabilities and disruptions as they ripple through production, rather than discovering shortages only after first-tier suppliers are affected.