Welcome to Chapter 9 of Flow.
Some highlights that I took from my few days in Birmingham
At Multimodal this week, I attended many presentations and panel discussions, met some great people, and walked the exhibition floor a number of times.
Attendance was good - not verified by any figures, just in discussions with regular attendees.
England were lucky to beat Ghana. Good to see a lot of stands turn on the match for kickoff!
The three words / phrases that I heard the most were:
AI
Sustainability / Decarbonisation
Digitisation of the industry
AI: The Chasm Between Rhetoric and Real-World Throughput
With the AI rhetoric - I think a lot of it was just wanting to look up to date and FOMO. When I dug deeper on the exhibition floor, real AI input was limited. I do think that the ones getting it right are the ones that are starting with the problem statement, and then going to the tech, rather than the other way around.
This boots-on-the-ground hunch is completely backed up by macro-level data. I’ve been reading the newly released State of AI in Supply Chain 2026 report independently surveyed by The Loadstar and supported by Raft. Their findings show that while the appetite is huge, deep operational deployment remains the exception, not the norm — only 22.2% of supply chain organizations have tangibly deployed AI at scale, and a mere 7.3% consider it core to their daily operations and decision-making.
To understand why, we have to look closely at where and how the technology is actually finding a foothold today:
The Low-Hanging Fruit: Document ingestion and extraction has emerged as one of the clearest early wins, with 79.7% of surveyed operators citing it as an area where AI has driven tangible operational impact. Supply chain runs on a vast, non-standardized flow of paperwork, emails, and PDFs, and automating that data ingestion fixes a massive headache. Among the companies deploying AI here, 52.6% reported direct, measurable improvements in internal collaboration across previously siloed departments.
Dynamic Exception Management: While reading documents is step one, the true frontier for competitive advantage is handling exceptions when shipments inevitably deviate from the plan. Right now, roughly 15.9% of freight forwarders and 15.6% of shippers are leveraging AI to triage operational exceptions. This is where AI stands to outpace traditional SaaS products; legacy software is built on a standardized model that struggles with custom edge cases, whereas AI can handle exception-heavy workflows according to the specific operational logic of your business.
The ROI Paradox: The biggest takeaway from my discussions at the Multimodal Conferences - and one that is heavily validated by the research data - is that the industry is fundamentally struggling to calculate the commercial value of AI. A staggering 62.8% of organizations admit they either haven't measured the financial ROI of their AI investments or are entirely unsure how to do it.
The friction stems from a flawed baseline expectation. While 89.5% of operators report clear speed and productivity gains from AI, more than half (56.4%) say it has reduced actual employee desk time by under 15% or not at all. This tells us that the standard metric of "hours saved" is missing the point. AI's value in a complex logistics ecosystem doesn't immediately show up as dropped headcount; it manifests as throughput scaled. It means an existing team can suddenly monitor more shipments, process more documents, and triage far more exceptions.
Sustainability and Decarbonisation
Sustainability and Decarbonisation were definitely at the forefront of most discussions, but there is an obvious dearth of real and deeper understanding in the field. Education is definitely required here, as outlined by the panel below. There is still very much a feeling that sustainability will cost us money, when the opposite can very much be the case, when a holistic view is taken. That has to be the next step - and again, predictive AI is beginning to play a big part here.
Digitisation and Dashboard Fatigue
Digitisation of intermodal terminals, again, was on everyone’s lips. But so were integration risk and dashboard fatigue. Operators definitely DO NOT want another dashboard. So integration becomes even more key.
Again, the data mirrors this perfectly. Nearly half (48.7%) of the industry identifies integration with existing legacy tech stacks as a massive barrier to adoption. In fact, 65.8% of the industry leaders surveyed state that data quality and integration will be the absolute baseline that differentiates the leaders from the laggards over the next two to three years.
Ultimately, the barrier to scaling these systems isn't a lack of technical sophistication in the models; it is a human and structural bottleneck. Across the market, 53.8% of leaders state that a lack of in-house AI expertise and change-management capability is what keeps them stuck in limited pilots. Trust remains heavily guarded: only 4.3% of teams have high enough confidence to let AI Agents run autonomously, meaning a "human-in-the-loop" review posture (67.9%) remains the absolute operational standard.
The tech solutions that win out won't be the ones forcing operators to look at another disconnected screen. The true advantage will come from building the internal operating models, the cleaner data connections, and the frontline trust required to turn isolated software experiments into repeatable, embedded processes.
Thanks for reading, and please feel free to share with anyone you feel may be interested.
Derek.
