Less human intensive, more data driven

Gangadhar Gude, Founder & CEO, atai.ai

AI provides transformational opportunity for logistics industry by improving customer experience, operational efficiency, faster turnaround time and lower cost while ensuring security and transparency. Macro environment requires industry to transform to be less human intensive, agile and data driven, all of which can be accelerated by AI adoption, shares Gangadhar Gude, Founder & CEO, atai.ai.

 What does Artificial Intelligence (AI) mean to logistics industry?

AI to logistics industry simply means how can I run my business more efficiently, how do I provide a better customer experience, how do I shorten the transaction time and how do I predict my business with more confidence? Logistics industry encompasses many areas including packaging, transportation, warehousing and last mile distribution of goods. Each area is very complex involving humans, equipment, space and time. Added to this complexity, the entire logistics industry is very fragmented with each area being handled by different players. Lack of information and data flow in a standardized and structured manner is a challenge found in this industry.

Technology adoption and innovation in manufacturing industries and changing customer expectations due to technology penetration in their day to day lives has put lot of pressure on the logistics industry to deliver the goods in shorter time at lower cost, while ensuring commitments, security and transparency. Artificial Intelligence powered by computer vision, process refinement and task automation can help this industry to truly reinvent itself. Additionally, AI can help logistics industry to become agile, so that companies can adapt to fast changing disruptions in macro-­‐economic circumstances pertaining to global trade and shifts in supply chain.

What best practice examples of AI used in other industries can be applied to logistics?

Let me give you three examples in the areas of planning, transportation and storing from our experience with various customers. One of the manufacturing companies in FMCG segment had an ERP system implemented which meticulously provides a consolidated sales projection month wise. This system was used by the plant managers to plan their monthly production. The manufacturing decision, raw material inventory planning, etc were being made meticulously using the data captured by the ERP system. However, the problem was under or over production resulting in revenue loss, customer complaints and enhanced friction between sale and operations team. It was a case of decisions turning bad due to uni-dimensional view of the data. AI in this case was used to understand the data and derive the true demand.

An AI based production planning system not only takes the demand requirements as captured by the ERP system, but also looks at the historical patterns and trends in sale projections and actual sales, takes into considerations external triggers/events, individual biases, pattern of goods movement (ex: goods with sufficient shelf life vs before expiry dates), etc. A comprehensive system will also analyse the true sales pattern, interpret the reason for a sale and also provide recommendations for sales strategy. The true power of AI is realized when this holistic view is taken.

Coming to this particular case in point, the operations planning system powered by AI helped the plant to move on to Just in time production without dumping the inventory on to the regional sales team and while meeting the demand of sales team. It also provided an evidence-based sales and operations interaction leading to smoother coordination.

The biggest issue in transportation is improper loading, tracking of the vehicle, gating of the vehicle and wait time or idle time of the transportation vehicle. Computer vision-based AI systems can monitor the process of packaging the goods and the loading or stuffing process. It can provide real-time alerts when an improper handling or exception is noticed. We have implemented this system which also ensures the good quality packages are loaded, guides the order of loading and placement in the vehicle based on the order of unloading point and nature of the package. It also provides tracks and manages the quantity loaded in a real-time, monitors the final covering or sealing process and automatically interacts with the ERP system to update the inventory and create the required paperwork. In this process, it also provides an evidence-based status to various stakeholders and authorities.

AI plays a major role in just-in-time vehicle dispatching taking into consideration the route planning, traffic conditions, environment conditions, regulatory conditions, external triggers (ex: possessions, unusual traffic activity, etc) and the readiness of the destination point. The AI enabled vehicle management system helps reduce the idle time and improves the utilization by up to 90% and 95% respectively. Similarly, AI is used in performing the vehicle survey and automatically gating in the vehicle thus reducing the wait time to near zero.

Artificial Intelligence can be used in the warehouse or yard planning, stacking and ingress/egress operations. It is typically used to derive the most optimal warehouse or yard strategy based on the Ingress and Egress patterns. Ex: AI has been used to define the location of the goods in the warehouse shelf or container in the yard based on various parameters like FIFO or LIFO for warehouses or zoning or special stow requirements for the containers. Similarly, AI also considers the optimization of equipment usage while determining the location of the goods or containers in the warehouse or yards. This helps reduce the equipment usage cost per operation and it can go as high as 30-50% based on our practical experience. Coming to equipment usage optimization, AI can be used to assign jobs to the equipment based on various factors like its proximity, capability and cost per usage parameters.

How can the logistics industry start using AI and take advantage of it? Where to begin?

Step1: Identify the key goals or purpose.

This exercise needs to be done at a business level and clear purpose needs to be defined. This acts like a guiding star in the journey of AI adoption as well as helps set the direction for the cultural shift in the organization.

Step 2: Perform a thorough independent study of the existing “As-­‐Is” process and methodology.

This helps us understand and appreciate the current status. In our opinion, this is one of the toughest parts of the entire exercise. Apart from understanding the as-is conditions, it also acts like a myth buster for many operations and procedures. We recommend a strong champion for this exercise and with a clear backing of the top management.

Step 3: Creation of heat map for solutions and process changes vis‐à‐vis the identified goals.

This helps in creating the right prioritization and performing a true ROI model for this initiative.

Step 4: Create a pilot project for technology demonstration

This step helps the organization get a feel of the real benefits of AI when the rubber hits the road. The pilot needs to be selected carefully as it provides (a) confidence to the management as well as various impacting stakeholders on the technology (b) helps understand the unstated requirements from various stake holders.

Step 5: Execute the program in well-defined phases

We strongly recommend against a big bang approach which can disrupt or confuse the larger teams and the current business. This stage should also be utilized to reorient select teams and train them. They can bring in very good inputs to ensure quality of the product. The solution should be tested in real deployment scenario. This experience is also utilized to prepare rollout training material.

Step 6: Planned rollout

This is a solution rollout phase and needs to be preceded by the training program to all the impacting the stake holders.

Step 7: Continuous Learning

Unlike other technologies, AI based solutions undergo continuous evolution by self learning from changing environment and conditions, it involves continuous data monitoring & fine tuning of AI models.

What are the opportunities and challenges in adapting AI? (in warehousing/trucking/port operations/vessel operations)?

Traditionally, this industry has not been a leader in technology adoption. When it comes to Indian & South-­‐East Asian market, the industry lags behind the global logistics market. Also, this industry is very fragmented which limits the organizational capacity for transformation programs. AI adoption cannot be looked at as an isolated solution as it requires deep process transformation and change management. To realize full benefits of AI, a lot of foundational work in digitization and automation may be required at many stages. Uncertain global environment may make it difficult for business leaders to approve new budgets.

Can AI be used in back office, operational, and customer relation management?

Logistics industries, while focusing on adopting AI in the core operational segment, companies can derive additional benefits by adopting it to support functions, back office, sales & marketing, human resource and finance.

A typical logistics transaction involves multi-level, manifold paperwork, multi-party contracts. Natural Language Processing and Computer Vision based AI integrated with Robotic Process Automation can be deployed to automate the documentation process to eliminate inefficiencies. AI also helps improve customer relationship by providing personalised customer support & experience using customer assist bots. Management can do operational planning & decision making using predictive analytics and recommendations using AI algorithms.