News / Flexport-Freightmate case tests ownership of AI freight data and workflows

Photo: © Bakhtiar Zein | Dreamstime.com

The legal battle between Flexport and freight-tech start-up Freightmate is increasingly becoming a test case for a question facing logistics tech companies: in the age of AI, is the most valuable intellectual property the code itself, or the data and workflows behind it?

While the case began as a trade secrets dispute, involving former employees allegedly taking company information to launch a rival business, recent court filings show the fight has shifted towards freight datasets, AI prompts, cloud infrastructure, and the records generated during AI product development.

The latest development came this month when Flexport filed a sanctions motion accusing Freightmate and its founders of failing to preserve key electronic evidence.

Flexport alleges that former employees Yingwei (Jason) Zhao and Bryan Lacaillade downloaded large quantities of company information before leaving to establish Freightmate, which included thousands of shipment-linked documents associated with Flexport’s internal freight forwarding platform.

According to court filings, many of those documents were later deleted from Freightmate systems. Flexport now argues that potentially relevant records relating to Google Cloud Platform resources, ChatGPT activity, AI prompts, and development workflows were also not adequately preserved after litigation became reasonably foreseeable.

The company contends that the missing records could have helped determine whether Flexport information was used in the development of Freightmate’s Docmate platform, an AI-powered system designed to automate the processing and validation of shipping documents.

Freightmate disputes the allegations, and has argued that Flexport materials were removed as part of a “clean room” process. It says Flexport has already been given access to its Git repository and source code history. According to Freightmate’s filings, review of the codebase has not identified any copying of Flexport code.

The dispute has increasingly centred on forensic evidence rather than software itself. Flexport is seeking access to cloud-based development records and other digital artefacts it believes could show how Docmate was built, while Freightmate argues many of the requests are unnecessary, given the discovery already provided.

And the outcome could have implications beyond the parties involved. As logistics companies invest heavily in AI tools capable of classifying shipping documents, extracting data, and automating operational processes, courts are likely to see more disputes over whether proprietary datasets, prompts, and automation workflows qualify for the same legal protections traditionally afforded to source code.

The next key milestone will be next month, when the court is due to hear Flexport’s sanctions motion.

A ruling could determine not only what evidence remains in play, but also provide an early indication of how courts may treat AI development records, datasets, and prompts in future logistics technology disputes.