What is a digital worker, really? (And how do you get one?)

By Barry Mitchell

Not too long ago, the term “digital worker” described a human employee with high digital skills. That was before the market defined it as a category of automated co-workers trained to do specific job tasks or execute business processes alongside their human colleagues.

Unfortunately, there’s no single, industry-wide definition of digital worker, which makes the concept harder to grasp and the potential impact on your business harder to forecast. There is some consensus among major robotic process automation (RPA) vendors when it comes to the concept of digital workforce: software-based labor that can perform specific tasks. But that definition is limited when you look at what’s available and what’s to come, technologically.

Digital worker defined

From the IBM Automation perspective, a digital worker is software-based labor that can independently execute meaningful parts of complex, end-to-end processes using multiple skills — skills that go beyond what RPA tools alone can offer. For example, a cash application digital worker may be able to autonomously perform parts of three traditional job roles — customer service representative, billing agent and cash applicator or dispute resolver — to perform an Order to Cash (OTC) process.

A digital worker represents labor, and labor requires skills. In this context, a skill is the application of one or more base capabilities, such as vision, natural language understanding or decision making. That skill is used to execute a specific job task, such as prioritizing work or preparing a report. Think of how a skilled radiologist or physician uses two base capabilities — vision and analytical reasoning (and years of experience) — to interpret an MRI image.

In sum, digital workers are multiskilled so they can execute across a reasonably useful scope of work and interact with humans to perform a sequence of tasks and activities that form a complete workflow. For example, you wouldn’t hire an administrator who could book meetings but couldn’t respond to emails. In the same vein, you wouldn’t want to create, say, an SAP administration digital worker who could monitor SAP but couldn’t create a ticket in Service Now or fix an issue it finds.

How do you get a digital worker?

You can go to a bot store and download what some RPA vendors call a digital worker, but these are still mostly simple, single task bots that can open a spreadsheet, download some data and post it into an ERP system. They don’t combine capabilities, such as document ingestion, natural language understanding or reasoning, from across the automation spectrum into the skills needed to execute end-to-end processes, such as OTC or Talent Acquisition (TA). Most bots aren’t true digital workers.

To help catapult productivity, most companies will need true digital workers to elevate and scale customer and employee experiences. And to get one, you’ll need to build it, get someone to build it for you or use digital workers from service providers “as a service.”

Below is an example of a digital worker IBM Automation Services created to support multiple clients in our business process outsourcing practice. 

Meet Ocash

Ocash is a digital cash application specialist: the latest recruit for the finance and accounting function. It’s often helpful to consider and position your digital workers in the roles they fill within your enterprise’s operations as shown in Figure 1 below.

Graphic representing the digital workforce persona using Ocash

Figure 1. Example of how to position a digital worker in the role it fills

To create Ocash, IBM's services team began with the workflow outcome in mind. Breaking down OTC into component parts, the team then focused on one of the more manual parts of the process. Within the cash application process, tasks were identified that could be automated, augmented and those that would still require a human to execute. IBM designed the process so Ocash could execute tasks best suited for automation and then call upon its human colleague(s) when needed.

In Figure 2 below, see some of the skills and associated tasks of the OTC digital worker. Ultimately, this could amount to more than 100 individual skills as the digital worker is taught and learns more.

Graphic representing skills and associated tasks of the Order to Cash digital worker

Figure 2. Digital worker skills — cash application

See an OTC digital worker in action (03:44).

Best practices for creating digital workers

When designing and building digital workers, four important design principles prevail:

1. Think big and take the view that the technology could run large parts of some processes

Where humans are involved should represent the proportion of work that deals with more complex exceptions, escalations and interactions — and improves the customer experience. Train bots to find and flag exceptions for human teammates, freeing humans from mundane monitoring activities.

2. Take an end-to-end view

Regularly step back and look at how all work tasks connect in terms of data flows, dependencies and impact on straight-through processing. Something you can automate at the start of the process may radically change what happens downstream. For example, automating the verification of invoice accuracy with a customer at the start of the process can significantly eliminate the number of payment disputes later on.

3. Build on your RPA beginnings

Digital workers are more than RPA, but that doesn’t mean they have to be cutting edge. The use of machine learning requires a lot of data and at least some idea, or hypothesis, for how you want to improve the process. Neither the data nor hypothesis may be available until the digital worker runs for a while, so don’t ignore the technologies that continue to serve you well.

Keep it simple to start: You can make the robot and the process smarter by enabling intelligent data capture, or you can help the digital worker make decisions — to know what to do next — by adding complex business rule engines.

4. It’s a people thing

You must consider the human element. Take a persona-based view. Enterprise design thinking can help you determine how you want humans to interact with digital workers to enable the intelligent workflow.


One of the key tenets of IBM’s future of work point-of-view is that we need to rethink how work gets done. In the new workforce, true digital workers will be working seamlessly with their human counterparts to get work done and deliver exceptional experiences to customers and employees. Digital workers can help take the robot out of the employees, freeing up your most important resource, your humans, to do the most important work. The key to digital workers at scale is to make them configurable and adaptable, in the same way our human workforce has been for decades.

Note: IBM Automation Services can provide prebuilt digital workers as a service, or help you purpose-build digital workers. And, in October, the IBM automation platform team will announce new software capabilities that include guardrails and tooling for building your own digital workers. We'll dive into the new AI and automation capabilities in the next issue.

Reclaim productive hours. Learn about the new intelligent digital worker tool for reconfiguring how work gets done.

Sign up for the IBM Automation Insider

What automation opportunities are out there? How do you get results?
Every other month our experts share five pieces of strategic content to help you drive growth through automation.