News / AI is needed in logistics – but it can be a multiplier or a liability

After years of digital transformation, freight forwarding still runs on a “surprisingly manual operating model”, according to a report from Bremen-based logistics AI company Deep Current. 

It says logistics firms have invested in operational systems, automation tools, dashboards and AI pilots over the past decade, but core workflows still depend heavily on emails, spreadsheets, PDFs, and fragmented systems. 

Deep Current surveyed logistics companies across Europe and the Middle East and found 72% planned to invest in document automation over the next 12 to 18 months, yet only 29% had implemented digital tools across core operational workflows. 

The report also found 57% of companies had experienced shipment delays caused by document errors, while 61% still relied on emails and spreadsheets for operational communication. Some 47% cited legacy system integration as the biggest barrier to adoption. 

“Digital maturity in 2026 will depend less on owning technology and more on connecting it,” said Tamim Fannoush, founder and CEO of Deep Current. 

The report noted that most logistics companies still operated across unconnected systems including TMS, ERP, and WMS platforms, while much of the real operational work was done by email, PDF, spreadsheet and messaging threads. Employees therefore become the ‘integration layer’, manually interpreting and transferring information between systems. 

“What’s interesting is that the friction is no longer really at the visibility layer. Most operators can now ‘see’ problems. The breakdown is happening at execution, where teams still have to manually interpret, validate, and move information between fragmented systems,” it explained.  

This is a problem for AI adoption, as Deep Current explained that critical freight inputs, such as bills of lading, invoices, customs declarations, contracts, and operational emails, were often ignored by AI pilots because they were messy and unstructured. 

“AI is not limited by model capability anymore, rather it is limited by data readiness. If your documents and communication streams are not structured and connected, your AI will never move beyond surface-level automation,” said Mr Fannoush. 

The report also argued that many AI projects failed because they were treated as add-ons. A new dashboard may be introduced, but the operator still has to download documents, check fields, update systems, investigate discrepancies, and respond to customers manually. 

Instead, Deep Current notes, AI becomes useful when it is embedded into the workflow itself. Then, incoming emails are interpreted automatically, documents are validated in real time, discrepancies are flagged early, and system updates happen without repeated manual entry. 

For freight forwarders, the stakes are practical. A missing HS code, incorrect consignee detail, or invoice mismatch can delay an entire shipment. Embedded AI, the report argued, could shift document handling from reactive checking to continuous validation. 

The company also suggested that the industry’s long pursuit of visibility was no longer enough. Dashboards and control towers can show something has gone wrong, but they do not necessarily help teams decide what to do next. 

“Visibility tells you there is a problem, while decision intelligence tells you what to do about it. In 2026, the competitive edge comes from shortening the time between signal and action,” explained Mr Fannoush. 

“Digitally mature logistics companies” are likely to focus more on decision intelligence, predictive resilience, and governance. Deep Current explained that that meant using AI to prioritise exceptions, recommend actions, model scenarios, and identify risks across routes, suppliers, and tradelanes before disruption hit. 

The report also warned that automation needed clear ownership and boundaries. As AI begins to influence shipment routing, document validation, and customer communication, companies will need rules for when systems can act, when humans must intervene, and how decisions are audited. 

“Operational AI is not about replacing logistics professionals; it is about elevating them. Governance and skills will determine whether AI becomes a multiplier or a liability,” said Mr Fannoush.