Reference architecture for an enterprise agentic landscape using ServiceNow AI Agent Studio, n8n, and MCP to automate PepsiCo logistics incident resolution; presented at ServiceNow World Forum NYC 2025 during RiseUp's "The Future of Agentic AI in Enterprise Organizations" panel.
PepsiCo’s job is to make sure their products, like Gatorade or Lay’s, reach retailers on time and in the right quantities.
Their entire business can really be summed up in one phrase I learned from Prashant Arora, Director of Strategy and Transformation at PepsiCo, during ServiceNow’s Knowledge 2025:
Someone has to step in fast: find another route, set a new ETA, and update Whole Foods so their shelves don’t sit empty.
Meanwhile, Schneider, our logistics partner, needs us to dispatch an alternate delivery route so that our delivery SLAs are still being met.
To address challenges like this, Berlynsea and I built closed-loop orchestration layers comprised of AI Agents that allows operations teams to quickly analyze the financial impact of a breakdown, identify the best rerouting options, and coordinate updates across logistics partners and retail clients in real time.
The result? Resolution processes that are faster, smarter, and more cost-efficient. This helps everyone involved and PepsiCo can keep creating more smiles with every sip and every bite!
It’s important to distinguish between MCP and A2A because they serve different layers of communication.
MCP focuses on translation and context sharing between systems, while A2A focuses on how agents actually talk and collaborate.
Together, these two layers form the backbone of agentic collaboration across PepsiCo’s infrastructure — connecting data, tools, and reasoning across every platform.
Now, imagine connecting SAP Joule, AWS Bedrock AgentCore, and ServiceNow AIA into a single intelligent ecosystem.
