Talking Supply Chain: Moving from AI pilot to execution with AWS’ Petra Schindler-Carter

Artificial intelligence has dominated supply chain conversations for years, but many organizations are still trying to determine where AI is delivering measurable operational value versus where it remains stuck in experimentation mode. In this episode of Talking Supply Chain, Petra Schindler-Carter, head of retail and consumer packaged goods at Amazon Web Services, joins host Brian Straight to discuss how leading retailers and CPG companies are moving AI beyond dashboards and pilots into real-world execution.

The conversation explores where AI is already driving measurable impact, including demand forecasting, inventory optimization, customer service automation, and exception management, and why foundational data accessibility, workflow integration, and organizational alignment remain critical barriers for many companies. Schindler-Carter also examines how companies like PepsiCo and adidas are using digital twins, AI-driven forecasting, and real-time customer demand signals to connect front-end commerce experiences with back-end supply chain decisions.

The discussion also dives into Agentic AI, digital twins, multi-cloud environments, autonomous decision-making, and why the future of supply chains will likely be “augmented” rather than fully autonomous.

Key themes and takeaways

AI is finally moving from experimentation into operational execution

For years, many organizations approached AI as a series of isolated pilot projects. Schindler-Carter argues that the companies now seeing measurable ROI are those embedding AI directly into core operating processes such as demand forecasting, inventory optimization, and customer service automation.

She notes that some AI applications are already becoming operationally mature, particularly in retail and CPG environments where speed, forecasting accuracy, and customer responsiveness directly impact margins.

Key takeaway: AI is delivering the greatest value when it is embedded into day-to-day operational workflows rather than treated as a standalone innovation project.

Key quote: “The companies that think from the beginning about ‘what decisions can we help automate with AI rather than saying, let’s just do an AI project,’ have a much better chance of succeeding and making a real business impact.”

Clean, accessible data remains the foundation of successful AI initiatives

One of the biggest barriers preventing organizations from scaling AI is not necessarily poor data quality, but inaccessible and fragmented data environments. Schindler-Carter emphasizes that organizations succeeding with AI have invested heavily in creating unified data models and real-time ingestion pipelines.

She argues that companies stuck in pilot mode often attempt to layer AI onto disconnected legacy systems without first addressing the underlying data architecture.

Key takeaway: The organizations scaling AI most effectively are prioritizing accessible, connected operational data before deploying advanced AI applications.

Key quote: “The biggest blocker that I see isn’t the data quality but the accessibility.”

Digital twins are evolving from visualization tools into operational decision engines

The discussion highlights how companies such as PepsiCo are using digital twins to simulate manufacturing plants, warehouse operations, and fulfillment processes before making physical operational changes.

Rather than simply visualizing operations through dashboards, AI-enabled digital twins allow organizations to model operational scenarios, identify bottlenecks, and autonomously optimize workflows in near real time.

Key takeaway: Digital twins are becoming operational intelligence platforms that help companies proactively optimize supply chain execution before disruptions occur.

Key quote: “What they’re doing is they’re creating high fidelity replicas of their manufacturing plants and their warehouses … and deploying AI agents that can test configuration changes virtually before they change anything physically.”

Multi-cloud and interoperability strategies are becoming essential

Schindler-Carter acknowledges that most Fortune 500 supply chains now operate across multiple cloud environments and legacy systems. Rather than forcing organizations into a single platform, she says modern AI architectures must support interoperability, APIs, and open data frameworks.

This flexibility is particularly important as companies attempt to connect ERP systems, warehouse operations, IoT sensors, transportation platforms, and customer demand data into unified decision environments.

Key takeaway: The future digital supply chain will rely on interoperable ecosystems rather than monolithic technology environments.

Key quote: “The way we approach these types of projects is interoperability over locking in.”

Agentic AI is becoming one of the most important developments in supply chain operations

One of the biggest emerging themes in the conversation centers on Agentic AI—systems capable of identifying disruptions, evaluating options, and autonomously taking action within predefined guardrails.

Schindler-Carter points to exception management and dynamic inventory repositioning as two of the most practical and immediately valuable applications of Agentic AI in supply chain environments.

Key takeaway: Agentic AI has the potential to significantly reduce operational latency by autonomously managing routine supply chain disruptions and inventory decisions.

Key quote: “The difference [is] the agent flagging a problem and the agent resolving it with defined guardrails.”

Retail leaders are connecting customer behavior directly to supply chain execution

Using adidas as an example, Schindler-Carter explains how advanced retailers are using AI to connect front-end customer interactions such as browsing behavior, promotions, and digital engagement directly into demand forecasting and inventory positioning systems.

This creates faster response cycles between customer demand shifts and operational supply chain decisions.

Key takeaway: AI is helping retailers unify customer experience data with supply chain execution in near real time.

Key quote: “Rather than looking in the rear-view mirror a few weeks later … they propagate those signals straight into inventory planning and production scheduling in near real time.”

Fully autonomous supply chains remain overhyped, for now

While autonomous supply chain marketing narratives continue gaining attention, Schindler-Carter believes the near-term future is more likely to involve “augmented” supply chains where AI handles operational scale and speed while humans continue overseeing strategy and governance.

She argues that AI-driven exception management and operational automation will become commonplace long before fully autonomous supply chains become reality.

Key takeaway: The most realistic near-term future is not fully autonomous supply chains, but AI-augmented operations combining machine speed with human judgment.

Key quote: “The future isn’t fully autonomous, but it’s augmented supply chains where the AI handles the volume, the speed, and the humans still handle the overall strategy and judgment.”

Why this conversation matters

As supply chains become more volatile, interconnected, and data-intensive, the conversation around AI is rapidly shifting from theoretical innovation to operational execution. Companies are no longer asking whether AI matters, they are asking where it delivers measurable business value, how to scale it responsibly, and how to integrate it into real-world workflows.

This episode provides a practical look at where AI is already transforming retail and CPG supply chains today and where the next wave of digital transformation is likely heading.

SC
MR