Logtech Essentials: From Procurement to Payment

Special Report

Logistics technology executives differ on AI coding threat

AI tools can increase logistics software coding capacity by between 20 % and 100 %. Shutterstock. com

By Eric Johnson

Earlier this year, a theory began to bubble up in technology circles as the use of AI-based coding tools started to take hold in virtually every corner of the software world: Will these tools destroy the software-as-a-service( SaaS) market?

Code generation tools such as Cursor, Windsurf, Claude, Devin, base44 and Lovable are already transforming the experience for software developers, allowing them to build products more quickly and add programming capacity to their teams without hiring expensive developers.

To be clear, the key benefit of these tools remains the augmentation of coding capacity, with logistics software executives telling the Journal of Commerce that they currently add anywhere from 20 % to 100 % more capacity to their teams.

But technology experts have also wondered if these coding tools might render existing software somewhat redundant. In such a scenario, a company that uses a piece of software could essentially build a new feature on top of that existing software, or, at an extreme level, they could theoretically build an entire product to replace that software.

Where this gets really interesting is if code generation software becomes powerful enough that non-technical people could build features or entire systems for themselves, merely through prompts. That activity, often referred to as“ vibe coding,” can essentially be described as coding with English through a program that turns prompts into code.

Think of an operations employee at a shipper prompting a code generation tool to create an application that builds a carrier scorecard by pulling information from its transportation management system( TMS), its visibility provider, its freight brokers and receiving facilities. Or, more significantly, think of a shipper having the ability to just create its own TMS through a series of prompts.

“ Lovable just released‘ agent mode,’ and I think it could build a TMS that rivals existing systems that freight brokers use,” Gabe Pankonin, CEO at less-than-truckload broker Rocket Shipping, wrote in a July post on X.“ The age of SaaS TMS platforms is coming to an end way quicker than I thought.”

Easier to build

Pankonin told the Journal of Commerce he expects a lot of companies in freight technology to be disrupted by customers building features via AI agents.

“ The SaaS message has always been,‘ Use our platform, and you don’ t have to build it yourself,’” he said.“ In the next three years, it’ s going to be easier to build it yourself than

implement someone else’ s technology. With Lovable and base44, if you know how to prompt, you’ re a designer, and then you can do some real damage. I tell it what I want, but I don’ t have to code it.”

“ In the next three years, it’ s going to be easier to build it yourself than implement someone else’ s technology.”

The idea of using natural language to build software might sound appealing to companies for a number of reasons. For one, the product would be tailored specifically to the needs of the company building it, reshaping the age-old build-versus-buy dilemma. For companies that invest in their ability to build tools for themselves, the speed of development could also be an advantage.

But it’ s not so simple, of course. The reality is that large companies— whether shippers, logistics companies or carriers— tend to rely on established enterprise software for a reason. It’ s stable, continually maintained and generally effective.

36 Journal of Commerce | November 3, 2025 www. joc. com