AI in Logistics: What Works and What Doesn't

The Role of AI in Logistics: What Works and What Doesn’t

Everybody is talking about AI, but following discussions with supply chain leaders in London last week it is clear that there are still a lot of questions about where to start and what it all means. The pressure to integrate smart technology solutions has never been higher, but AI is not a silver bullet; but rather another arrow in the quiver.

While some approaches deliver significant improvements in efficiency and cost reduction, others fail to meet expectations or even complicate operations.

So let's have a closer look at what does and what does not work when applying AI use cases to your supply chain.

Applications of AI in Logistics That Work

Predictive Analytics for Demand Forecasting

One of the most impactful uses of AI in logistics is predictive analytics, which enables companies to accurately forecast demand. By analysing historical data and identifying patterns, AI systems can predict future demand, helping businesses optimise inventory levels, waste and pricing.

Why It Works:

Predictive analytics driven by AI can process vast amounts of data, far more than any manual system. The ability to incorporate external data sources such as weather patterns, economic trends, and consumer behaviour leads to more accurate demand forecasts, allowing companies to streamline supply chains and respond more effectively to fluctuations.

Example: Greenscreens.ai is a dynamic pricing platform that uses artificial intelligence (AI) and machine learning to provide real-time, predictive pricing for freight and logistics companies. Its main focus is on helping businesses in the transportation and logistics industry to optimise their pricing strategies for spot market transactions, balancing profitability with competitiveness. and understanding the right price points when brokering freight.

Route Optimisation for Freight and Delivery

AI-powered route optimisation tools have transformed last-mile delivery and freight transportation. These systems use real-time data, including traffic conditions, weather, and road closures, to determine the most efficient delivery routes, reducing fuel costs and improving delivery times.

Why It Works:

AI can analyse multiple variables in real-time, enabling dynamic route adjustments that human planners simply can’t achieve at scale. This is especially valuable in urban environments where delivery windows are tight and traffic is unpredictable. AI-based route optimisation ensures that deliveries are not only on time but also cost-efficient, improving overall customer satisfaction.

Example: Companies like CAROZ THE TMS use AI to optimise delivery routes, cutting fuel consumption and reducing the number of kilometres driven, saving both time and money.

Warehouse Automation and Robotics

AI-powered robots and automation systems are now commonplace in modern warehouses. These systems manage tasks such as sorting, packing, and transporting goods within the warehouse. AI helps improve accuracy and efficiency, reducing human error and enabling around-the-clock operations.

Why It Works:

AI-driven warehouse automation integrates seamlessly with inventory management systems, reducing manual labor and speeding up operations. These systems are particularly effective in handling repetitive tasks and can scale during peak times, offering a flexible solution to manage varying workloads.

Example: Active Ants UK leverages AI to optimise its e-commerce fulfilment operations, integrating AI-driven robotic systems for tasks such as order picking, sorting, and packing within its warehouses.

Building Parcel Carrier Integrations

The logistics space relies heavily on interconnectivity to various partners, suppliers and carriers. For niche tasks, AI helps to analyse requirements, write code and automate the quality assurance process, delivering outputs in seconds when provided with the right prompts and guidance from engineers.

Why It Works:

AI-driven tasks need to be specific and niche. They don't work in isolation, but when combined with highly specialised human knowledge that provides the right guidance AI can automate a significant amount of the work involved with software engineering.

Example: Gluey utilises AI to build parcel carrier integrations in 2 days, rather than 6-12 weeks, which is common in the industry. This allows companies requiring the addition of new services to rapidly meet customer requests and scale into new markets, achieving significant revenue growth.

AI Applications in Logistics That Don’t Deliver

Over-Automation Without Human Oversight

While automation and AI are powerful, some companies have failed by over-automating processes without considering the need for human oversight. Fully autonomous decision-making without human input can lead to mistakes that AI isn't equipped to handle, such as exceptions or unforeseen circumstances in the supply chain.

Why It Fails:

AI is effective at handling repetitive tasks and large data sets, but it struggles with nuance and context in real-world scenarios. When logistics companies automate too much without integrating human decision-making at critical points, errors can go unnoticed, leading to operational disruptions.

Example: Some early adopters of fully autonomous warehouse systems experienced increased downtime when robots were unable to handle complex or unexpected situations, resulting in costly interruptions to operations.

AI Systems That Lack Integration with Legacy Software

Many logistics companies are still using legacy systems to manage their supply chains. AI tools that fail to integrate with these systems can create more problems than they solve, adding complexity instead of improving efficiency.

Why It Fails:

Without seamless integration, AI systems can create data silos, causing breakdowns in communication across different departments or parts of the supply chain. Incompatible systems lead to inefficiencies, higher costs, and frustration for teams trying to manage operations across disconnected platforms.

Example: A logistics provider might invest in an AI-powered fleet management tool, but if it doesn’t communicate with the company’s existing order management system, the data remains fragmented, leading to suboptimal decision-making.

Misaligned AI Expectations

Many companies invest in AI without a clear understanding of what it can realistically achieve. AI is a powerful tool, but it is not a magical solution to fix everything. Businesses that expect immediate, revolutionary changes without accounting for proper training, implementation time, and necessary human intervention often see disappointing results.

Why It Fails:

AI implementations that aren’t grounded in realistic expectations can lead to over-investment and underperformance. AI solutions need time to be fine-tuned, and companies must invest in the right training and talent to manage these systems effectively.

Example: Some logistics companies invest heavily in AI-driven predictive analytics but are disappointed when initial forecasts are inaccurate, often due to a lack of data or proper setup. These systems require fine-tuning and iteration to deliver value.

Key Takeaways: What Makes AI Work in Logistics

1. Clear Objectives: Companies that define specific goals for AI (e.g., reducing delivery times, optimising inventory) are more likely to see success than those that adopt AI without clear outcomes in mind.
2. Integration with Human Expertise: AI works best when it complements, rather than replaces, human expertise. For example, AI can suggest optimal routes, but human drivers might still need to adapt to real-time challenges.
3. Data Quality: The effectiveness of AI systems relies heavily on the quality and volume of data they have access to. Companies that clean and integrate their data before implementing AI tend to achieve better results, but remember that data is seldom perfect, so find the right balance here.
4. Scalability: Successful AI applications are those that scale efficiently across the supply chain, from forecasting to warehouse management to last-mile delivery. Starting small and scaling up ensures the systems are adaptable and effective.

Conclusion

AI offers tremendous potential to revolutionise the logistics industry, but its success depends on careful implementation, integration, and realistic expectations. The most effective AI solutions enhance decision-making, improve efficiency, and offer flexibility, while those that fail often do so because of over-automation, poor integration, or misaligned expectations. By focusing on specific applications that are grounded in practical use cases and complement human expertise, logistics companies can harness the true power of AI to stay competitive in a fast-evolving market.

*AI tools were utilised in putting this article together.

Thanks to Matthew Silver , Khalil Ashong CMILT and Maurits Jongens