AI Implementation: Lessons from 2025’s Experience

A recent Supply Chain Management Review article takes a look at what pundits got right – and wrong -about AI use in supply chains during 2025, and draws upon the lessons learned to make the following assessments about what companies need to do in order to ensure the success of their own AI initiatives during the current year:

Standardize before you automate. Clean, consistent data and aligned processes must come first. AI does not fix poor data hygiene, rather it exposes it at scale. Leading organizations are integrating AI in deliberate phases, starting with narrowly scoped, high-confidence use cases and progressing toward more advanced, agent-based workflows only after data, controls, and accountability are proven.

Use agents to support decisions, not replace accountability. 73% of executives expect AI agents to deliver a significant competitive edge. Clear ownership matters: leaders must define who approves AI-driven actions and who is accountable when automated recommendations introduce risk.

Treat AI as a core third-party risk domain. AI usage, data dependencies, and automated decision-making should be assessed alongside cybersecurity, privacy, compliance, and financial risk, especially when capabilities are delivered through vendors and service providers.

Extend visibility beyond the enterprise. Understanding internal AI usage is no longer sufficient. Risk leaders need insight into how both internal employees and suppliers deploy AI, how those systems evolve, and where automated decisions intersect with critical workflows. Without that visibility, scaling agent-based systems will only magnify blind spots.

Move from static assessments to continuous oversight. A quarter of executives say trust gaps are their biggest hurdle to ROI from AI, creating a need for more responsible implementation and oversight. Risk frameworks must adapt as internal and supplier AI usage changes, regulations evolve, and new data dependencies emerge, rather than relying solely on point-in-time reviews.

One of the big lessons from 2025 that’s reflected in these recommendations is that AI adoption outpaced the implementation of governance systems necessary to provide effective visibility, oversight, and accountability for AI tools.

The article says that organizations often struggled to track their own AI usages due to the problems of “Shadow AI” use resulting from product updates or new vendor relationships. That problem was exacerbated by the fact that visibility into vendor usage of AI was That challenge multiplied across supply chains. Visibility into how vendors were using AI, including data sources, training practices, and automated decision logic, remained limited or nonexistent.