Worldwide Express Automates LTL Freight Exceptions
Worldwide Express has modernized its less-than-truckload freight billing process using a multi-agent artificial intelligence solution built on AWS infrastructure.
Radwan AbdulhadiEditor and technology writer
The Challenge of LTL Freight Exceptions
Worldwide Express, operating as part of ShipStation Global, generates approximately $5 billion in systemwide revenue, serving more than 130,000 customers alongside over 75 less-than-truckload (LTL) carriers. However, this massive operational scale introduced a significant financial bottleneck. Every month, $15 million in free cash flow remained frozen due to unresolved billing exceptions that required manual tracking across different carrier rules.
In the LTL freight sector, roughly 14 percent of invoices fail standard three-way matching between purchase orders, goods receipts, and carrier invoices. When a match fails, the invoice enters a manual exception-handling pipeline categorized into line-item mismatches, pricing variances, and unexpected accessorial charges. Each dispute traditionally required a five-step manual review process that consumed substantial bandwidth from the FinOps team.
Deploying a Multi-Agent Solution
To resolve these operational hurdles, Worldwide Express collaborated with AWS and LTM to design a multi-agentic AI solution that integrates directly into their transportation management system, SpeedShip. The project moved from assessment to production in less than six months following an engagement with the AWS Business Transformation & Value Advisory team in June 2025.
Rather than relying on rigid rule-based process automation, which struggles with the contextual judgment needed for variable freight tariffs, the company turned to advanced artificial intelligence tools. The underlying architecture leverages Amazon Bedrock to provide foundational AI reasoning capabilities, while system execution relies on Amazon Elastic Compute Cloud (Amazon EC2).
Specialized AI Agents in Action
When an invoice exception is flagged, the system routes the data to one of three specialized AI agents designed to handle specific dispute categories. The Invoice Matching Agent identifies discrepancies in quantities and service codes against shipment records. Meanwhile, the Pricing Variance Agent evaluates carrier charges against contracted rates, and the Accessorial Validation Agent checks unexpected fees against actual delivery events.
Each custom agent operates using a defined instruction prompt, a set of actions that integrate with the SpeedShip transportation management system, and access to specific carrier knowledge bases. Before making a recommendation, the agent retrieves the relevant carrier tariff rules and standard operating procedures. Developers can review regional availability and foundation models using the documentation for Supported models by AWS Region in Amazon Bedrock.
Integration and Workflow Control
The custom agents connect to SpeedShip via the same APIs utilized by human analysts, allowing the AI to evaluate shipment details, delivery proof, and contract rates within familiar workflows. Because every carrier maintains distinct billing rules and dispute procedures, the system stores individual carrier documentation so that every decision remains grounded in specific operating standards without requiring separate codebases for each carrier.
During the initial rollout, human reviewers utilized the SpeedShip user interface to approve recommendations before finalization. Once system accuracy exceeded 99 percent, bulk approvals became the default workflow. Dashboards track automation rates, error categories, resolution times, and human override frequencies to continuously refine prompt instructions based on historical feedback.
Measurable Business Results
Six months after launching the automated exception management workflow across more than 40 LTL carriers, Worldwide Express reported substantial financial and operational gains. The specialized agents currently close approximately 10,000 exceptions weekly, representing nearly 50 percent of total volume.
The initiative successfully reduced active exceptions by 75 percent, accelerated time-to-cash by four weeks, generated $1 million in annual savings, and delivered a one-time $10 million free cash flow infusion. Organizations interested in exploring similar modernization frameworks can request a conversation through official AWS channels.
Sources
- AWS Machine Learning BlogHow Worldwide Express automated freight exceptions with Amazon Bedrock
Continue chronologically
Related entity coverage
Related stories
ShipStation Global Builds AI IT Help Desk on Bedrock
ShipStation Global built Deckhand, an AI IT help desk agent on Amazon Bedrock AgentCore that resolves channel queries and files Jira tickets.
Building Agentic Retrieval on Amazon Bedrock with AWS CDK
Learn how to build enterprise agentic retrieval on a Managed Amazon Bedrock Knowledge Base using the AWS CDK and Amazon Bedrock AgentCore.
Amazon Bedrock AgentCore Automates PowerPoint Generation
Learn how the new Amazon Bedrock AgentCore AI agent system automates PowerPoint presentation creation from enterprise documents in under 90 seconds.
Circle K Cuts DevOps Incident Triage Times With Amazon Bedrock
Circle K's DevOps team accelerated incident triage from up to 60 minutes down to under 15 minutes by implementing an agentic AI system on Amazon Bedrock.
