API automation is extending the division of labour to routine office work

Automation typically does not replace an entire job, but instead automates certain repetitious aspects of that job. According to McKinsey & Co, the available technology could enable the automation of at least 30% of activities being performed in approximately 60% of jobs. API Workflow Automation is one tool being used to automate parts of work performed in a typical office setting.

Automation divides jobs into tasks, not workers out of them

The idea of dividing work into multiple tasks, so each task can be done more efficiently, looking at the totality of the work being performed, is one of the oldest concepts in economics. Adam Smith had described in one of his early works (Wealth of Nations) a factory producing pins, in which approximately ten workers employed in various stages of production, performing just one activity at a time (e.g., drawing the wire, then cutting it), were able to produce up to approximately forty-eight thousand pins a day. In contrast, the same factory performing the same work by only one worker would produce fewer than twenty pins per day. API Automation is taking the same concept and applying it to the area of knowledge-based work. When working with a customer onboarding process, for example, break the total of that process down into smaller discrete sequential activities (e.g., creating a record in the CRM, sending a welcome email, invoicing for the first time) and assign those sequential activities based on which systems could best complete each activity. In this way, the employee remains employed performing those activities, but only the repetitive pieces or segments of their job have been automated. The principle behind this model is that specialisation results in the lowest economic cost per unit of output.

Specialisation works because each task falls to whoever gives up the least

Through specialisation, participants in the division of labour focus their abilities on producing a higher dollar value with less time and money invested in any one specific task. The rationale for this is:

• That the opportunity cost of performing a task should fall to the person who will incur a smaller opportunity cost in taking on that task.
• That automation allows machines to never tire of repetitive, monotonous tasks. Therefore, the opportunity cost for a machine to perform repetitive work is effectively zero.
• That the employee who is skilled in an area incurs a much higher opportunity cost for doing routine tasks than a machine. All of the employee's time spent on doing research and development, writing proposals, performing general services, and providing customer service is tied to the hours that he or she is not performing that task.

When an organisation assigns its routines to machines, it frees its employees to perform those judgement-based tasks that machines still have difficulty handling. As a result, the overall productivity of the organisation has been increased. The teller at the bank provides an example of this principle. Automated teller machines have assumed the counting and dispensing of money. However, the presence of an ATM has not resulted in the disappearance of tellers, but rather has resulted in a move by tellers to perform service-oriented roles that the ATM is incapable of fulfilling. Therefore, the function of the teller has been automated, rather than the individual teller.

The old barrier was coordination, not the labour itself

The main reason for not splitting out routine tasks is that coordinating between the parts of an organisation was cost-prohibitive in the past. Historically, businesses would have kept routine tasks and activities grouped together within one person because the cost of coordinating the transfer of information (i.e., formatting, checking, delivery, etc.) was not worth it for the business. Low-code platforms break down the historical barrier to separating routine and judgement-based tasks by linking applications via API nodes that enable users to capture leads in one application and then have them transferred, formatted correctly, into three other applications without needing to be retyped. The more cost-effective the process of coordination becomes, the finer a division of labour will be within an organisation.

As the economies of scale of a product increase, so too will the associated costs for that product decrease. Creating an effective workflow for a process may require substantial investment, but the cost associated with running that same workflow an additional thousand times is close to nothing. Therefore, as the quantity of output increases, the more the average cost per unit of product will be reduced. If a finance department develops an automatic invoice-matching system, it will incur the costs of developing that product only the first time it is used. The cost of processing every subsequent invoice will be negligible, as more invoices are processed using this automated system.

Therefore, the principles illustrated by the pin factory applying to the creation of products are the same as those found with the development of automated API functions that will allow the tasks of humans and software to be reallocated, while allowing the human element to continue to perform the judgement-based tasks that require human input.