Merlyn Consult / Transport & Supply Chain Insights
AI and Machine Learning Are Transforming the Transport Sector
The smartest transport operation is not necessarily the one with the most technology. It is the one that makes better decisions, earlier — and gives planners, carriers and drivers the tools to act on them.
01 / The reality in 2026
Beyond the hype. Into daily operations.
Transport and warehousing are rich in operational data: orders, vehicles, routes, delivery windows, warehouse scans, temperature readings and carrier performance. The challenge is turning that data into useful decisions — without making an already busy operation more complicated.
Artificial intelligence is a broad term for systems that carry out tasks such as prediction, classification and recommendation. Machine learning is one way to build such systems: models learn patterns from data rather than following only fixed instructions. Not every useful optimisation feature needs AI, and that distinction matters when evaluating software.
of shippers in Descartes’ 2026 benchmark said they were using AI at scale — far fewer than those exploring or piloting it.
of respondents identified routing, transport or shipment optimisation among leading AI use cases.
Data quality was the leading barrier to scaling AI, cited by 33% of shippers and 45% of logistics service providers.
Source: Descartes 2026 Transportation Management Benchmark Survey. Survey findings describe respondents, not the whole industry.
02 / A more connected journey
Four decisions that shape every delivery
Rather than treating AI as one giant project, follow a shipment through the operation. Each stage offers a realistic point of improvement.
03 / Where the value lies
Six use cases worth understanding
Your own trucks — or the right external carrier?
Compare vehicle availability, shipment requirements, rates, service history and capacity. AI-assisted recommendations can help dispatchers choose between own fleet and subcontracted transport, while retaining control over commercial and service decisions.
Practical outcome: more informed allocation decisions.
Make the route respond to reality
Traffic, time windows, loading constraints and changing delivery priorities rarely stay still. Predictive models and optimisation algorithms help identify alternative routes and anticipate delays. The best solution balances cost, service and emissions rather than simply choosing the shortest distance.
Practical outcome: fewer avoidable kilometres and surprises.
Protect the cold chain before it breaks
Connected sensors already provide temperature alerts. Predictive analytics can go further by identifying patterns linked to refrigeration faults, route exposure or handling risks. AI supports earlier intervention, but does not replace validated temperature monitoring and regulatory controls.
Practical outcome: better exception handling and product protection.
Give planners clarity, not another dashboard
A visual planning board becomes more useful when recommended assignments, capacity conflicts and late-arrival risks appear alongside the schedule. Planners need to understand why a recommendation was made — and be able to override it when local knowledge matters.
Practical outcome: faster, explainable decisions.
Close the loop between plan and road
Driver apps and onboard systems connect live status, proof of delivery, navigation and messages with the transport management system. Machine learning can use that information to improve estimated arrival times and flag exceptions, provided the data is reliable and privacy is respected.
Practical outcome: better communication and more useful ETAs.
Synchronise the inbound and outbound flow
At a busy cross-dock, minutes matter. Forecasting inbound arrivals and coordinating dock doors, handling capacity and outbound departures can reduce waiting and unnecessary movement. Integration between TMS, WMS and yard operations is often more important than a standalone AI tool.
Practical outcome: smoother throughput and less rehandling.
What about autonomous trucks? They remain important to watch, but operating safely at scale is considerably more complex than improving transport planning. For most SMEs, decision support and process integration are more accessible places to start. The OECD’s 2026 mobility analysis similarly describes autonomous operation as an area with substantial technical and regulatory challenges.
04 / A practical approach
How to start without turning AI into an IT experiment
The priority is not to buy an AI platform. It is to solve a worthwhile operational problem and establish whether the available data can support a better decision.
Choose one operational pain point
Examples: unreliable ETAs, manual carrier selection or avoidable empty running. Define the process owner and business impact.
Check the data foundations
Inspect data completeness, timestamps, interfaces and ownership across TMS, WMS, telematics and ERP.
Run a bounded pilot
Compare a baseline with measurable results: planning time, on-time delivery, kilometres, exceptions or cost per shipment.
Keep people in control
Make recommendations transparent, set escalation rules and train planners and drivers before scaling.
Better data + clearer processes + empowered people = a stronger starting point than AI hype.
05 / The important caveats
Efficiency is valuable. Reliability is essential.
- Data quality: missing scans, delayed telematics feeds and inconsistent master data can undermine even sophisticated models.
- Integration: confirm how a proposed tool works with existing TMS, WMS, ERP and carrier connections before committing.
- Security and privacy: protect shipment, customer and driver data; define access and retention rules.
- Accountability: keep safety-critical and commercially sensitive decisions subject to appropriate human oversight.
- Regulation: assess the applicable EU AI Act requirements based on the specific system and its intended use rather than assuming every transport application is classified alike.
For EU regulatory dates and obligations, consult the European Commission AI Act Service Desk.
06 / The Merlyn perspective
Less fascination with technology. More focus on decisions.
AI and machine learning are changing transport and logistics, but not through one dramatic leap. The progress that matters is often incremental: more reliable ETAs, better carrier choices, earlier warnings, faster planning and more coordinated flows between transport and warehouse operations.
For SMEs in particular, the soundest strategy is to start with business processes, identify an achievable use case and select technology independently. AI should strengthen a well-run operation — not disguise one that still needs fixing.
