The AI dialog in logistics has shifted. on s. Within the newest installment of Lean Quarterly Dive with BigRig and the crew at Lean Options Group (LSG), Thomas Wasson sat down with Alfonso Quijano, chief know-how officer and co-founder at LSG, to speak about how AI and automation are literally enjoying out inside freight brokerages and logistics corporations.
For a lot of 2025, frontier AI labs promised agentic workflows would remake complete job classes. Quijano stated that promise didn’t land the way in which it was pitched, and firms with out present know-how infrastructure are discovering the hole between entry to AI and the power to make use of it properly.
“I’m seeing that the experimentation section is usually over,” Quijano stated. “You tried it, you went into it with out numerous expertise. Firms that didn’t have know-how groups discovered themselves investing a ton of cash into AI to see if it labored for them.”
Broad entry to AI platforms didn’t eradicate the necessity for fundamentals, although.
“Even when the know-how had democratized entry and all people may obtain OpenAI and Claude, it doesn’t imply that you simply’re an AI firm,” Quijano stated. “You really want good mission administration. You want good change administration. You could have good management of your prices. Who knew?”
The rise and fallout of “vibe coding” (utilizing AI to generate software program with minimal conventional improvement) was a cautionary story for the business, in line with Quijano.
“An unbiased writer put out that 99% of vibe coded apps are within the rubbish,” Quijano stated. “[These apps] don’t make any cash. They get revealed after which get unpublished from the infrastructure platform.”
The lesson, he stated, is that velocity of creation doesn’t equate to enterprise worth.
“Rubbish in, rubbish out. Even when it’s quicker to create know-how, that doesn’t imply that you simply’re going to create extra companies,” Quijano stated.
In keeping with Quijano, purposes constructed swiftly in the course of the top of the AI wave have been typically fragile as soon as deployed. “It appears to work, however in actuality it’s very brittle,” he stated. “While you apply it in actual life and in manufacturing, it will possibly trigger some fairly vital harm inside what you are promoting.”
Slightly than chasing AI-first branding, Quijano argued the extra mature method treats AI as one element inside a broader automation technique, deployed solely the place it provides worth.
“It’s higher for AI to be current however not talked about than the opposite manner round,” Quijano stated. “You type of must make it invisible for it to be adopted because it ought to inside organizations.”
Many corporations have wrapped AI round issues that when had easy, rules-based options.
“You’re taking one thing that might have in any other case been fairly easy and now utilizing AI for it, and the unpredictable nature of the result is inflicting points,” Quijano stated. “You’re type of higher off saying, ‘I’m going to convey automation into my firm’ quite than being an AI-first or AI-native answer. That automation can have AI elements when it’s needed, however not more than that.”
In keeping with Quijano, organizations ought to conceptually deal with the know-how not as software program to configure, however as personnel to onboard. “I don’t suppose it’s a software. I feel it’s an worker,” he stated. “What do you do with a junior worker that joins your organization? You prepare them. You make sure that you give them a really outlined job description in order that they know precisely what they should do.”
Meaning documented processes, exception dealing with and ongoing coaching.
“You want a really outlined SOP,” Quijano stated. “You could be sure that your course of is well-documented and that you simply account for potential errors within the course of. Exception administration. You could have steady coaching. You could shut the loop.”
Lean has been constructing that closed-loop infrastructure immediately into its personal know-how stack. The corporate just lately expanded its LeanTek platform with new AI governance, workforce intelligence and value visibility instruments, giving organizations execution-level value monitoring, workflow suggestions loops and centralized governance dashboards designed to maneuver AI deployments previous experimentation and into accountable, measurable manufacturing use.
“Organizations are more and more targeted on tips on how to handle, measure and scale AI responsibly,” Quijano stated of the replace. “These new capabilities give leaders the boldness to scale AI whereas making certain transparency, accountability and measurable enterprise outcomes.”
Skipping the groundwork produces the identical consequence as a mismanaged rent.
“You’re ultimately going to have to fireplace this new worker, which is one other factor that I’m seeing very, very regularly now,” Quijano stated. “There was a wave of AI tasks that began earlier in ‘25 that corporations are instantly discovering methods to get out of.”
Quijano additionally pushed again on the concept that AI can repair damaged processes or underperforming groups. “You possibly can truly improve the output of errors into your system, probably,” he stated.
LSG was based when brokerages wanted nearshore expertise to fill roles they couldn’t employees domestically. The situations at the moment are such that there’s an elevated demand for individuals who sit on the intersection of operations, enterprise, and know-how.
“You might have to have the ability to transfer between understanding what the correct metrics are that can help you present that this actually is offering worth for the enterprise, understanding tips on how to implement it, and understanding what know-how implementation actually seems like,” Quijano stated.
That shift requires rethinking present roles, Quijano stated, describing a “cross-pollination” method borrowed from software program improvement. “You might have very sturdy know-how groups that generally you wish to convert into pods which have larger protection capability amongst different tasks which are adjoining to your personal,” he stated.
Many corporations deploying AI at scale are grappling with runaway token spend. In keeping with main AI corporations, there are a whole bunch of 1000’s of brokers deployed throughout varied corporations. Quijano says that many corporations are constructing greater than they want, and that they’re rapidly counting on token utilization greater than anticipated.
LSG is itself reliant on Claude for engineering work.
“We’re huge proponents of an AI-first technique in direction of coding, however each main AI firm can also be saying it’s important to begin fascinated with your personal AI fashions,” Quijano stated. “I feel that’s the subsequent path that that is going to go.”
It’s questionable whether or not the latest frontier fashions justify their added value. “While you evaluate [the newest models] to what you’ve the aptitude to do in the present day, it’s about six to seven p.c higher,” Quijano stated. “However on the similar time, it’s twice as costly. In my eyes this appears extra like a technique to monetize and pump this AI narrative quite than truly a humongous stride in AI enchancment. We’re undoubtedly within the legislation of diminishing returns proper now.”
That plateau raises a much bigger query, he stated. “Does that imply that we’re at some extent the place the open supply neighborhood can present me higher bang for my buck for deploying AI fashions than what the frontier labs are going to do? It’s but to be seen. We’re doing it, but it surely’s not simple. Now it requires know-how information, infrastructure. You require DevOps expertise. Now you’re again to the fundamentals.”
LSG’s technique has centered from the beginning on augmenting folks quite than displacing them. Quijano expects the remainder of the business to converge towards that mannequin because the hype cycle settles.
“We determined to do this from the get-go,” he stated. “We speak about human plus AI, or tech-powered people. We speak about elevating the position of the human to have extra obligations.” That elevation requires actual funding in reskilling the present workforce, he stated. “New obligations contain coaching and the upskilling of those roles in order that they are often extra tech-focused. They will have the power to not solely use the software program, however be part of it, a part of the answer.”
Click on right here to study extra about Lean Options Group.
The publish The AI Experimentation Part Is Over appeared first on BigRig.