Stampfree needed help improving their computer vision models. Ortom helped them do this, updating the model architecture, collecting training data, and deploying the models on improved infrastructure.

Challenge

StampFree is transforming the parcel delivery industry by replacing printed labels with simple handwritten codes. Tim Higginbotham, CIO at StampFree, explains: "Our mission is about customer convenience and being more environmentally friendly. 3bn labels are produced every day, involving printers, ink, glue and of course paper. The environmental impact is significant, but also the way that customers have to deal with multiple carriers, multiple retailers—we thought there was a better way of doing it."

The customer-facing side of the solution worked well. Writing a code on a parcel and taking a photo proved straightforward and convenient. However, StampFree faced a critical technical challenge on the carrier side. Major carriers like Royal Mail, Evri, and InPost operate high-volume sorting facilities designed entirely around scanning barcodes. For StampFree's solution to scale, carriers needed to read handwritten codes as easily as barcodes—and do so at speed across hundreds or thousands of parcels in automated production lines.

The company's existing AI system, developed years earlier, struggled with specific scenarios. "The detection side was definitely an area of weakness for us," Tim notes. "We were not able to deal with text that was anything but level, so if it was wonky or upside down, sideways, it just wouldn't read it." For high-volume carrier operations requiring accuracy, speed, and reliability, these limitations posed a significant barrier to market adoption.

When StampFree's original AI developer became unavailable, Tim needed to find a specialist partner who could enhance the existing computer vision system. Through Scottish Enterprise, who were supporting StampFree's development through grant funding, Tim connected with Ortom.

Solution

Ortom approached the project as a strategic technology partner, beginning with a thorough assessment of StampFree's existing codebase. Understanding how the previous system worked—and precisely where it was failing—was essential before implementing improvements.

The solution required addressing multiple technical challenges simultaneously. The first priority was improving text detection, particularly for rotated, angled, or inverted handwriting. Parcels arrive at sorting facilities in various orientations, and the system needed to handle this variability reliably. Beyond detection, the overall recognition accuracy needed enhancement to meet carrier operational requirements.

A critical innovation came through introducing AWS Mechanical Turk for data collection. Machine learning systems require substantial training data, and StampFree had been manually collecting handwriting samples—a time-consuming process that limited the volume and diversity of data available. Mechanical Turk enabled crowdsourced collection of handwritten codes from workers worldwide, dramatically expanding the training dataset.

"Your suggestion around using the AWS Mechanical Turk was fantastic," Tim recalls. "That made a significant difference. We weren't aware of that service. And suddenly to be able to crowdsource testing data was brilliant." Ortom developed a custom tool to streamline the data collection process, ensuring quality and efficiency.

The work extended beyond the core recognition improvements. Ortom recommended migrating StampFree's infrastructure to AWS SageMaker, a specialized AI server environment, improving both performance and operational efficiency. Throughout the project, Ortom collaborated closely with StampFree's existing development team in India, maintaining clear communication through weekly calls and working sessions.

"Your suggestion around using AWS Mechanical Turk was fantastic. That made a significant difference... suddenly to be able to crowdsource testing data was brilliant. And the little tool you created to enable that to go really smoothly—that was fantastic."

Tim Higginbotham
CIO, StampFree

Results

• Enhanced Text Detection Capabilities: The improved AI system now reliably handles rotated, angled, and inverted handwriting—a critical requirement for automated carrier sorting facilities where parcel orientation varies.

• Improved Recognition Accuracy: Enhanced training data and updated algorithms delivered measurably better handwriting recognition performance across diverse writing styles and conditions.

• Enabled Live Carrier Trials: The improved system provided StampFree with confidence to proceed with live operational trials with a European carrier, validating the solution in real-world, high-volume environments.

• Streamlined Data Collection Process: The Mechanical Turk integration and custom tooling established an ongoing capability for StampFree to continuously improve their models with fresh training data.

• Strengthened Technical Infrastructure: Migration to AWS SageMaker provided a more robust, scalable platform for the AI system, supporting future growth and operational demands.

• Positioned for Market Expansion: With enhanced technical capabilities, StampFree gained greater confidence presenting their solution to carriers and expanding into additional markets beyond their existing deployments.

"It gives us a lot more confidence in how we present the solution and in its operational capability. We are running live trials now with a European carrier to validate the process."

Tim Higginbotham
CIO, StampFree

Conclusion

StampFree now possesses a significantly enhanced AI system capable of meeting the demanding requirements of carrier operations. The improved handwriting recognition technology is supporting live trials with European carriers and strengthening the company's position as it expands its label-free delivery solution across additional markets.

This project demonstrates Ortom's collaborative approach to working with innovative technology companies. We combined technical expertise in computer vision and machine learning with practical solutions like streamlined data collection, infrastructure improvements, and close collaboration with existing development teams. By focusing on the specific operational requirements of StampFree's carrier partners, we helped remove a critical barrier to scaling their environmentally focused innovation.

"Let's be honest—typically, stereotype developers like to work on their own. Being able to work with yourself, the communication was great, and the style was very open, I thought was fantastic. That's a rare combination... I would absolutely recommend working with Ortom to other businesses."

Tim Higginbotham
CIO, StampFree