How Worldwide Express automated freight exceptions with Amazon Bedrock

Worldwide Express, a major player in the transportation industry, has streamlined its billing exception management process by leveraging a multi-agent AI solution built on Amazon Bedrock and AWS infrastructure. This innovative approach has brought significant improvements in operational efficiency and financial results.

The company, valued at approximately $5 billion, operates with over 75 less-than-truckload carriers and serves more than 130,000 customers. However, the sheer scale of its operations had led to a considerable problem: $15 million in free cash flow was being frozen in unresolved billing exceptions each month. Resolving these exceptions required a labor-intensive manual process, with each exception requiring individual attention from a dedicated team of specialists.

To address this issue, Worldwide Express partnered with AWS and LTM to develop a sophisticated AI-driven solution. The solution was designed to automate the exception resolution process, reducing the time and resources required to address billing discrepancies. Within six months of implementation, the company reported a remarkable 75 percent reduction in active exceptions, a 4-week improvement in time-to-cash, and a significant $1 million annual savings.

The key to this success lies in the multi-agentic AI framework that underpins the solution. The system is composed of three specialized agents, each handling a specific type of billing dispute: Invoice Matching Agent, Pricing Variance Agent, and Accessorial Validation Agent. These agents are designed to work in tandem, pulling relevant carrier billing rules and policies before making recommendations.

Each agent is equipped with a unique instruction prompt that defines its role, decision boundaries, and output format. They also have access to carrier-specific knowledge bases, which include tariff rules, surcharge definitions, and dispute procedures. This ensures that the AI's decisions are grounded in the exact policies and procedures of each carrier, minimizing the risk of errors and inconsistencies.

The agents communicate with SpeedShip, the company's transportation management system, through familiar APIs. This integration allows the AI to see exactly what a human reviewer would see, including contract rates, shipment details, and delivery proof. The AI then generates a resolution within the same workflow, following the same steps that a human reviewer would take.

During the initial rollout, a human reviewer was involved in the approval process, using the SpeedShip UI to validate the AI's recommendations before they were finalized. However, as the AI's accuracy approached 99 percent, the company transitioned to a bulk approval model, significantly speeding up the exception resolution process.

Overall, the implementation of this multi-agentic AI solution has transformed Worldwide Express's exception management process, freeing up valuable resources and enabling the company to focus on generating cash flow from orders rather than chasing down billing disputes. This innovative approach has positioned Worldwide Express as a leader in operational efficiency and financial management within the industry.