The power of artificial intelligence in warehouse management

Introduction

It is predicted that at least 15% of daily work decisions will be made autonomously through Agentic AI by 2028 [1], with the global market forecasted to grow at an astonishing CAGR (Compound Annual Growth Rate) of 46.2% until 2030 [2]. 

The emergence of Agentic AI has the potential to reshape almost every industry as we know it. But for warehousing it is the catalyst needed to transform intelligent warehouses (that use AI automation based on human defined rules) to adaptive warehouses (where agents autonomously learn, optimise and act without human intervention). 

‘Logistics is no longer the stable sector it used to be,’ highlights Chris Coote, Director of Product at Dexory. ‘The explosion of e-commerce, changes in consumer demand and labour shortages are triggering dynamic loads in the system and warehouses need to respond to maintain supply chain resilience.’

‘The key to achieving this is improving efficiency and therefore questioning how to better utilise resources,’ adds Coote. ‘This all points towards autonomy, whether that is using autonomous robots to collect inventory data faster, AI solutions to draw out actionable insights, or self-optimising AI agents for logistics that better orchestrate overall warehouse operations.’

‍

AI vs Agentic AI – what’s the difference?

Artificial Intelligence (AI) has been around since the 1950s, working in the background of well-known software tools, such as Matlab, for decades. However, today’s availability of more powerful, yet cheaper computing power has facilitated the use of AI at scale, which is why it has suddenly flooded our everyday lives. 

AI essentially simulates human intelligence and is defined as machines that use algorithms to ingest and analyse training data to establish trends and predict future states. Consequently, AI can execute tasks, but is ultimately based on equations, data and constraints defined by humans. 

Agentic AI on the other hand are machines capable of thinking and acting autonomously and independently to meet human-defined goals. They ingest real time data from a variety of sources and automatically adapt to changing conditions, following new rules without humans in the loop. Multiple AI agents can work together, forming a network that synthesises tasks and workflows, optimising operations.

‘AI agents are essentially goal-seeking bots,’ says Coote. ‘They thrive on being given specific tasks and then working out the best course of action. For example, if tasked with finding the best slot in the warehouse for a particular product, the AI agent would first analyse all the information at its disposal. It would try to understand what slots are free, if the product needs to be stored in a specific area for more efficient picking, whether it is suitable for storing high up in the racks or at a certain temperature or humidity.’

‘It uses this data to form an opinion of the best outcome according to pre-defined goals and then takes action,’ continues Coote. ‘Instead of having one agent responsible for major tasks, the industry is tending towards deploying ecosystems of AI agents, where each agent is responsible for smaller tasks but communicates with others to solve large-scale activities.’

‍

‍

Advantages of utilising agentic AI

The latest studies show that in general, Agentic AI systems significantly outperform traditional AI techniques, reducing task completion time by 34.2% whilst increasing accuracy by 7.7% and improving resource utilisation by 13.6% [3].

Faster task completion

Traditional workflows rely on tasks being completed sequentially which means waiting for one task to be signed off, before the next begins, accumulating delays. Whereas AI agents execute multiple tasks simultaneously, accelerating cycle times.

More responsive workflows

AI agents are continuously ingesting the latest data in real time. By utilising this information they can reorder priorities, identify anomalies before they turn into issues and adjust staffing schedules based on predicted demand; proactively adapting to changing conditions. 

Boosting resilience

The endless adaptability of AI agents allows them to keep processes running, regardless of any issues that may arise. This ensures the supply chain remains resilient despite disruptions.

‍

Trusting AI

With less human intervention than ever before, it’s natural to question whether the outcomes of AI and AI agents can be fully trusted. ‘AI is just another tool and like any tool, to get reliable results you have to provide it with accurate and representative data,’ highlights Richard Williams, Senior Robotics Engineer at Dexory. ‘That’s why the 99.9% accuracy achieved by our autonomous inventory scanning robots is a key enabler for the move to Agentic AI.’

Article: Dexory’s next-gen Autonomous Mobile Robot sensors for ultimate accuracy.

‘You then need to validate how the AI interprets this data with comprehensive and well-trained models,’ adds Williams. ‘When it comes to joining multiple AI agents together, you also need to employ standard methods of communication, such as MCP [Model Context Protocol], to ensure information isn’t misinterpreted before the AI makes a decision.’

The final stage involves verifying the results from AI through rigorous testing programmes in both digital simulation and real-world environments. ‘To further reinforce trust, we may also have to use explainable AI throughout the early stages of adoption,’ says Coote. ‘By showing the workings of how the AI calculated a decision along with quantifying the real-world impact of that decision, customers will find it much easier to trust the outcomes of AI agents.’

‍

‍

Will Agentic AI replace employees?

Understandably, the biggest concern for businesses moving towards a more AI-led future is the social impact on employees. However, AI should be viewed as a tool that helps amplify human capabilities. ‘We are not trying to replace staff or develop completely automated warehouses with no humans involved,’ concludes Williams. ‘Instead, we are trying to free staff from repetitive, manual tasks so that they can focus on higher-level activities and decision making.’

‘We can then support this decision making process with an intelligent AI-driven platform that analyses copious amounts of data and surfaces intuitive insights,’ continues Williams. ‘This enables staff to make the best decisions, with the most amount of information, within the shortest time possible.’

‍

References

[1] G.A., 2024. Gartner Top 10 Strategic Technology Trends for 2025 [Online]. Gartner.

[2] Enterprise Agentic AI Market (2025 - 2030) [Online]. Grand View Research.

[3] P.D.S., 2025. Agentic AI: A Quantitative Analysis of Performance and Applications [Online]. Journal of Advances in Artificial Intelligence.