ePost Global (EPG) operates a worldwide logistics network, serving retailers and e-commerce businesses in more than 200 countries and territories. Its customer service representatives (CSRs) are on the front line, handling a high volume of customer inquiries every day. From April through June 2025, the team received and processed 14,145 customer emails, averaging 155 per day across the entire CSR team.

Of the 83% of emails that were shipping-related, the analysis revealed several key patterns:

• 47.3% required nothing more than a tracking update.

• 19% stalled because of missing information, leading to unnecessary back-and-forth.

• 42% raised issues that required speedy responses to maintain the highest level of customer satisfaction.

The existing workflow was manual and time-intensive. CSRs spent hours every day triaging and replying to repetitive inquiries, leaving less time for complex cases that required personalized attention.

The limits of the manual process

On a typical day, EPG’s CSRs could keep pace with the email volume—barely—but scaling was impossible. The patterns identified in the analysis underscored the problem:

• Volumes could grow quickly, creating bottlenecks for the team.

• CSRs spent valuable time chasing missing information, which slowed response times and frustrated customers.

• Customer trust was at risk when negative inquiries weren’t addressed quickly enough to meet expectations.

The data showed that 79% of cases were resolved in two or three emails. But each exchange consumed valuable time that could have been avoided if the right information and answers were delivered up front.

A request for automation

EPG asked Tenjumps whether AI could help transform their CSR workflow. During the discovery phase, Tenjumps’ strategic and engineering teams analyzed the 14,145 emails and 5,104 tracking numbers provided by EPG. The goal was to identify high-frequency inquiry patterns and design a solution that would ease the burden on human agents without disrupting the existing systems.

The findings informed a clear roadmap for a custom AI-driven solution that would slot into EPG’s workflow and move quickly into development.

Building the CSR AI Assistant – JumpAssist

Within two months, Tenjumps developed a phase-one minimum viable product (MVP) deployment: an AI-powered chatbot called JumpAssist, tailored to EPG’s needs. The model was trained on actual customer emails to recognize sentiment, identify intent, and deliver instant responses for the most common query categories, particularly tracking and status requests.

The solution included:

• Discovery and Ideation: Tenjumps partnered with EPG to analyze goals and pain points, using the three-month email study to surface patterns and opportunities for automation.

• Research and Design: The team mapped user journeys, defined success metrics, and designed chatbot workflows to ensure smooth integration with EPG’s existing systems.

• AI and Machine Learning: Tenjumps built a custom chatbot trained to classify shipping-related inquiries, detect sentiment, and provide instant answers.

• Data Engineering and Analytics: The team analyzed more than 14,000 real customer emails to build the training set and created reporting tools to track volume, resolution rates, and customer sentiment.

• Consulting and Strategy: Tenjumps developed a clear roadmap for deployment, focused on measurable efficiency gains and minimal lift for EPG’s internal team.

JumpAssist was integrated into EPG’s customer support website. When visitors interacted with it, the bot would notice if relevant information was missing from the requests and ask for it up front. It provided immediate answers where possible, created a support ticket if the customer requested it, and escalated only the more complex cases to human CSRs.

How JumpAssist worked

At the core was a natural language model that scanned incoming emails and matched them to high-frequency response templates.

• If a customer asked about a tracking number, the chatbot could provide a status update instantly.

• If required data was missing, the chatbot prompted the sender for the exact information needed, reducing delays.

• Neutral- and negative-sentiment cases could be flagged for faster attention by CSRs.

The system was designed for scalability, operating 24/7 and capable of handling thousands of inquiries without additional staffing.

The results CSRs could see immediately

The outcomes were measurable and immediate:

• JumpAssist resolved more than 60% of daily tickets instantly, with no human intervention required.

• Human CSRs were freed to focus on more complicated cases requiring judgment or escalation.

• Average resolution time dropped from multiple email exchanges to a single automated reply in many cases.

The faster responses and reduced back-and-forth immediately improved the customer experience, setting the stage for higher customer satisfaction. 

This deployment was framed as phase one, validating the chatbot’s ability to handle high-frequency email categories. Future enhancements will expand functionality, including skill-based routing, deeper sentiment analysis, and potential multichannel integration.

The business impact

Global logistics generates endless repeat inquiries that drain CSR time, while customers demand faster answers. This project demonstrates how a data-driven AI solution can deliver both speed in customer response times and scale in handling high-volume repeat inquiries with minimal CSR effort.

By starting with a three-month analysis of 14,145 real emails, Tenjumps built a working solution that immediately reduced workload and created measurable ROI. With future features already planned, the solution is positioned to become a core part of how EPG evolves the customer experience going forward.