A few years from now, will a company’s most important customer actually be a machine rather than a person? Although that might sound like science fiction right now, it could soon become reality thanks to continuous technological advancements. Machine customers, also known as custobots (customer robots), are algorithms that independently manage solutions and perform services. One example is refrigerators that independently replenish supplies when items are running low. Experts predict that the use of custobots will increase significantly in the coming years.
This development could prompt fundamental change in many areas of the economy, especially if automation and data-driven decisions are already established. According to research firm Gartner, there will be up to approximately 15 billion machine customers by 2028. Tobias Wölk, Product Manager at reichelt elektronik, is therefore confident that “Custobots will have a significant influence on trade and industry in the coming years.”
Increased efficiency and innovation: The rise of machine customers
There is a lot of future potential for machine customers and their influence on the economy. Advanced technologies like the Internet of Things, artificial intelligence and machine learning enable organisations not only to process large volumes of data more efficiently but also to use this data to predict market trends and customer needs more accurately. It is the knowledge acquired in this way that custobots use — their goal is to streamline operational processes and boost efficiency by interacting directly with suppliers and service providers and placing orders.
The difference between people and custobots is that the latter are guided more by logic and less by emotions. Their decisions are usually based solely on data and evaluations, which is why companies are being challenged to develop strategies for integrating and working with machine customers. Not only legal and technical considerations must be taken into account, but also any adjustments to operational processes.
Bound customers: A human sets the rules and the machine implements them
Roughly speaking, development of machine customers is divided into three distinct stages: bound customers, adaptable customers and autonomous customers. We are currently experiencing the first phase in the development of machine customers, and it can be observed in various services. During this stage of development, the custobots are bound to pre-programmed functions and follow certain rules. In this scenario the machine customers adopt the role of “co-customer”, taking on tasks assigned by people. These machines are referred to as “bound customers” because they operate within a clearly defined framework and adhere to predefined specifications.
One example of this is the HP Instant Ink printer, which monitors the ink level and orders supplies before the ink runs out. This is based on clear programming that specifies exactly when the machine should act and exactly what steps must then be followed. Services like Amazon Dash Replenishment are also a particularly good illustration of this. As an example, Amazon Dash Replenishment allows different devices, such as washing machines, to order laundry detergent at the touch of a button when supplies are running low.
“As a rule, integrating bound machine customers does not require any major adaptation for people”, explains Tobias Wölk. “For instance, if a company sets up automatic ordering for consumables that are regularly required, such as printer cartridges, this relieves the burden on office managers and buyers. All they need to do is set up the system once and then they can start benefiting from the automated workflows. The installation process is easy to understand and generally very quick. The integration of bound machine customers is a good starting point, especially for companies that are still at the start of their digitalisation and automation journey.”
Adaptable customers: A human sets the rules and AI acts independently
Even in the second phase of development, there is greater autonomy and decision-making freedom for machines. Although a human continues to set the rules, machine customers increasingly take independent action with the help of AI. Data and algorithms are used to make decisions. Human intervention is kept to a minimum.
Examples of this process are modern IoT devices that adapt their ordering strategies based on usage patterns and other variables. One example of how machine customers can be used in various household appliances is a refrigerator that automatically detects which products need to be purchased at regular intervals and reorders them as needed. The user defines in advance which products should be purchased and gives the order in advance to purchase those products when they run low. The robot detects the quantity of food in the refrigerator, independently initiates the ordering process and automatically selects the required product. What the robot is not yet able to do, however, is gather information about the user’s frequently purchased food items or favourite dishes and then use that information to purchase products for recipes that the user might also like to try. A task like this could only be achieved in the third phase of automation,
which marks a transition to greater autonomy and independence of the machines — although humans will continue to play a central role in defining the requirements. A solid data basis is important when it comes to successful training of the system. If a custobot is going to be used by an animal feed manufacturer to purchase ingredients, for instance, data about suppliers, prices and quantities of specific raw materials will provide valuable insight. However, the experience of the specialists working within the company is also indispensable when it comes to checking that the system is set up correctly, monitoring quality and making any adjustments that may be required. Likewise, if new animal feeds are created or the recipe is changed, the system must first learn the changes and it requires control. “So as we can see, a solid data basis is not the only important factor for successful integration of adaptable machine customers. It is also crucial that the specialists working with the system are extremely familiar with the possibilities and limits of the custobot and are able to supplement the process with their experience”, adds Product Manager Tobias Wölk.
Autonomous customers are essentially independent agents
In the third phase, custobots are able to act independently, making decisions based on the analysis and verification of large volumes of data. The human is no longer the main actor, but rather just points in the right direction. In this scenario the custobot acts independently on behalf of the human. Using algorithms, machine learning and artificial intelligence, machine customers are able to solve problems independently and complete complex tasks. Whereas requirements were previously specified by a human who then tasked a machine with taking some degree of independent action, it is now the case that a machine, based on the data, may recognise a requirement before the human and take action accordingly. Machines can also adapt more quickly to new information and make decisions in real time, which represents significant progress.
Autonomous vehicles that can have maintenance performed independently are a good example of this phase. The condition of the vehicle is monitored closely by a large number of sensors on the vehicle components. Orders are placed based on feedback from the sensors. This step can then be submitted to a human for approval, meaning that the final decision remains with a human.
Another example is intelligent logistics systems that manage and optimise stock levels using custobots. For example, a manufacturer of home electronics could use purchasing data to identify trends as they arise and to order the required components at an early stage and at the best terms. The orders are placed based on business data such as products purchased and sold, customer purchasing behaviour including the customer journey or searches on the website, the composition of shopping baskets, supplier price trends and much more. Even the slightest change in demand can alert the custobot to new trends. In addition, the system is able to anticipate potential supply bottlenecks and order the required goods early and at lower prices in order to avoid the bottleneck. “Human expertise and experience will still be required, in particular for monitoring the systems by means of random spot-checking — something that is particularly important in the event of unforeseen scenarios such as pandemics or wars”, says Product Manager Tobias Wölk. “However, the advances with big data and AI-based evaluation are showing us even now how autonomous custobots can be used to our advantage. They are fast, don’t need sleep or holidays and can make a whole host of decisions every day in a way that wouldn’t be possible for humans. This enables companies to achieve significant efficiency gains and a competitive advantage. However, a more complex supply chain is the result of this progress. Companies will need to become accustomed to frequently encountering actors that don’t play by “human” rules. Sooner or later, suppliers and wholesalers will start responding with their own automated systems, and competition will then ramp up once again.”
Future challenges
Forecasts show that machine customers will start playing a meaningful role in the economy in the next few years. Many companies will be focusing increasingly on automation and customer interactions. Thanks to big data, AI, machine learning and other technological advances, machine customer applications will start to open up in a variety of areas. This development will bring both great opportunity and significant risk. One significant risk is the scope of authority of the custobots and their access to sensitive and business-critical data. Data protection safeguards must therefore be implemented from the outset in order not to compromise confidential information. In addition, integration with blockchain technology is expected, and this could facilitate even more secure solutions.
Another risk is that, unlike humans, custobots make decisions on a purely rational basis. Human factors are generally not taken into account in decisions based on data, but the human factor and the inherent experience, intuition and empathy is often indispensable when it comes to making decisions that have far-reaching consequences. With advanced AI models and extended machine learning techniques, custobots could learn to take these factors into account over time, thus allowing them to make decisions that are even more accurate and effective.
“The future of custobots is promising but also challenging”, concludes Wölk. “As we look to the future, machine customers will not only have a huge impact on various sectors but will also play an active role in the lives of many people.”
Images: Adobe stock
