Augment your team, automate day-to-day work and elevate employees to focus on what matters most.

Covered in this chapter

  • What is a digital employee?
  • Digital employee example
  • Anatomy of a digital employee
  • Debunking common myths

The Art of Automation: Table of Contents

While there are many excellent books on digital workers and chatbots, for this Art of Automation book, we’ve decided to highlight an exciting new technology that provides capabilities that extends the best of chatbots, robotic process automation (RPA) and digital workers — enter digital employees.

In the fall of 2020, Allen Chan, CTO of IBM Business Automation, was roaming the virtual executive halls at IBM and socializing the idea of Digital Workers 2.0 to fuel the future of work. In parallel, I was tasked with bringing to market a new interactive artificial intelligence (AI) technology coming out of IBM Research labs that was codenamed Verdi. Verdi was an AI-based planning algorithm that intelligently orchestrated agents based on natural language instructions.

At the time, the automation market was reeling from the exponential growth in robotic process automation (RPA) and converging toward an integrated automation platform as inspired by Gartner’s hyperautomation concept. In parallel, the conversational AI trend had also reached a level of adoption. Conversational AIs, or chatbots, were now commonplace in help-desk and customer-support scenarios, offering a rich natural language interface to answer pre-trained questions and execute pre-determined actions. The next evolution of Conversational AI, predicted by Gartner, was to be multimodal, with deep context and with a wide range of enterprise app integrations to tackle more complex virtual assistant use cases.

These market trends did not wholistically capture the concept of a digital employee, but the technology landscape offered a level of maturity to set the foundation for IBM’s point of view on a hybrid workforce. IBM’s integrated, seamless automation platform (IBM Cloud Pak® for Business Automation), IBM Research’s interactive AI innovation (Verdi) and Allen Chan’s digital worker concept came together to form watsonx Orchestrate. As announced by IBM’s CEO, Arvind Krishna, at IBM Think 2020, watsonx Orchestrate empowers companies to compose their own digital employees to augment their human workforce and automate work when and where they need it.

What is a digital employee?

The best way to understand what a digital employee is to focus on the ’employee’ part first and then the ‘digital’. Meaning, let’s understand the job role and function of the digital employee and then focus on the technology that makes it possible. Critical to understanding the job role and function of a digital employee is to first understand the job role and function of the human employee it is going to support.

Let’s consider a real example.

Digital employees in human resources

Meet Pat

Pat is an HR Business Partner in a large enterprise. She is responsible for working with business units within her company to plan and execute strategic human resource goals. One of those goals is to ensure high employee engagement and retention through an effective employee promotion cycle.

Pat and her team have a quarterly promotional cycle. Each quarter, criteria for nomination are finalized in partnership with business leadership. Then the nomination process is kicked off with 100+ managers submitting their nominations back to HR. Once nominations are received, HR needs to aggregate the nominees into a master list, augment each nominee record with 120+ data points and then submit the nominees for approval to both business and HR leadership. 

Meet HiRo

HIRo is a digital HR Business Partner. HiRo is Pat’s new sidekick. Pat is still responsible for the overall quarterly promotional cycle, but HiRo is here do a lot of the leg work so that Pat can focus on more strategic conversations with her business stakeholders related to the promotion cycle.

HiRo interfaces with Pat and other employees using Slack, the chat platform deployed at Pat’s organization. That is, Pat and other employees are able to chat with HiRo using Slack.

Figure 1: HiRo’s digital employee profile.

HiRo has been built with key HR skills:

  • “Kickoff Email” skill: Every quarter, Pat tells HiRo to fire off the kickoff email to the 100+ managers. HiRo gets the latest promotion criteria and the due date for the nominations from Pat. Taking this information, HiRo populates a pre-defined email template and fires it off to the managers, letting them know when their nominations are due.
  • “Add to Nomination List” skill: When a manager selects an employee to nominate, HiRo adds the employee to the nomination list.
  • “Nomination List Approval” skill: Once all the nominations have been received, Pat tells HiRo to generate the Master Nomination spreadsheet. This skill not only merges the 100+ nomination lists received from the managers but also augments the employee data by pulling 120 data points from various HR systems (performance management, compensation, etc.).

HiRo also has skills to help employees perform key tasks in Workday and ADP. But just looking at the three skills above, HiRo is already saving Pat and every HR Business Partner in her team days of chasing managers, pulling and validating data and crunching spreadsheets. HiRo can pull employee data for each nominee from multiple HR systems in seconds, a task that would take HR days to complete.

Anatomy of a digital employee

A digital employee is a complex set of technologies that need to come together to form a whole:

Figure 2: The anatomy of a digital employee.

Planning: The key to intelligent orchestration

There are plenty of existing automation technologies that provide orchestration of work. Companies have deployed workflow and case management systems to orchestrate end-to-end business processes, such as processing applications, claims, support tickets, invoices, etc. The orchestration provides a standardized and, ideally, optimized workflow. This workflow, however, is predefined — each decision point and path are pre-determined. In real life, things change and don’t always fit into pre-determined scenarios. That’s where intelligent orchestration can fill the gap.

The key to intelligent orchestration is planning. A digital employee interfaces with the world around it through events. It interfaces with human employees through conversation and these trigger natural language events. It can also interface with business systems by listening to notifications when something changes — these trigger data events.

Given an event, the digital employee needs to find the best course of action to address the incoming event using a planner. At runtime, the planner evaluates the set of skills it has at its disposal and scores them based on how confident the skill is that it can address the incoming event. In the simple scenario above, a manager asks HiRo to pull an employee record. HiRo has a Workday skill to ‘get’ an employee record. This skill scores at >90%, letting the planner know it should be selected to handle the incoming event. But what if HiRo also has an ADP skill to ‘get’ employee payroll information (i.e., another type of employee record). This ADP skill would also score high. This is where the intelligence of the planner and conversational nature of the digital employee enable it to go back to the manager to clarify which type of employee record they need. The planner can also learn from this interaction and begin to weight one skill over another to help it optimize its selection algorithm for the future.

So, the planner can understand the incoming event and select the best skill. It can also disambiguate between similar skills. Additionally, it can learn and improve its scoring algorithm over time.

But this is just the beginning for intelligent orchestration. The planner also needs to be intelligent enough to sequence skills as needed to truly orchestrate a workflow on the fly. Once HiRo retrieves the employee record for the manager, the manager decides to add the employee to the promotion nomination list. The nomination list will be reviewed by upline management who needs key data points not available in the core employee record to decide if the employee nomination should be approved (data points such as time in current position, last salary increase, etc.). HiRo has the skills to retrieve this information from the various backend HR systems. When the manager asks HiRo to add the employee to the nomination list, the “Add to Nomination List” skill lets the planner know that it requires the following input parameters to execute the skill:

  • Core employee record
  • Time in current position
  • Last salary increase amount
  • Last salary increase date

The planner already has the core employee record but is missing the other parameters. It does have the ADP skill to get employee salary information, however. Here’s where the intelligence of the planning algorithm kicks into overdrive. It knows it needs to execute the “Add to Nomination List” skill, but it can sequence the ADP skill ahead of it in execution to get the missing input parameters. The planner does this by looking at all the skills it has at its disposal and determining which ones can output the values needed as input to the “Add to Nomination List” skill. Using this technique, you can see how the planner can sequence two skills, but it can also be extended to sequence three or more skills. This is true intelligent orchestration.

Understanding: Adds context to intelligence

Understanding goes hand-in-hand with planning. Understanding brings context to the intelligence. This gives the intelligence an understanding of the state of the world as it knows it today. 

It starts with foundationally knowing the self and its surroundings. For digital employees to function in an organization, they need to know where they fit into an organization — which team they belong to and who they work for. They have direct knowledge of the user directory and have their own record in the org tree. This gives the understanding of which employees they work with and which manager they report to so that they have a clear path of escalation when they need help. And digital employees do need help from time to time.

The second layer of understanding is knowing what objects the digital employee can work with. Each digital employee maintains a schema describing the objects it knows how to work with. In the example above, HiRo is built with the understanding that it will work with employee record objects. When HiRo retrieves the employee record for the manager, it can add it to its memory of objects. This enables the manager to work with HiRo over multiple interactions using that object. The manager can now ask HiRo to get more information on this employee (e.g., last year’s Performance Review). The manager can also tell HiRo to add this employee to the promotion nomination list. In either case, HiRo understands the context (i.e., the employee) of the conversation and completes the requests on behalf of the manager.

Identity: Gives purpose

Much like human employees, a digital employee exists to satisfy a specific job function — it has a specific purpose. Each digital employee has its own employee profile that personifies its purpose. For example, HiRo is a Digital HR Business Partner. When employees want to interact with it, they know what its purpose is, what it can do for them (e.g., the skills) and how they can interact with it (e.g., via its chat handle, email address, etc.).

With its identity, it also has its own credentials to the various business systems it needs to work with.  HiRo has access to Workday but has its own credentials so HiRo’s manager can control what actions and data HiRo can work with in Workday. If HiRo is meant to only serve the North America team, HiRo’s credentials can be limited in Workday to only work with NA employee data. These credentials not only give managers a way to control access for digital employees as they would their human employees, but also gives all employees a way to track what the digital employee has done in their business systems.

Skills: Get work done

Given all the intelligence and identity of the digital employee, though fairly smart, it is not very useful unless it has the ability to execute work. That’s where skills come in. Skills are the atomic unit of work that a digital employee can perform. 

Skills come in a variety of flavors:

  • Productivity skills: These enable the digital employee to work with common productivity tools like email and calendars to send emails and schedule events. When an employee works with a digital employee, they would not want to use a chat interface to craft an email; that’s hardly more efficient than using their email client directly. But if a manager asks HiRo to pull an employee record, they can easily ask HiRo to also email that record to a colleague for review. 
  • Analytical skills: Employees spend countless hours pulling data from various business systems to generate reports and glean insights. Many others don’t — even though they should — because its time consuming or they lack the technical skills. This category of skills enables employees to get the digital employee to query business data using natural language. The digital employee generates the SQL statement behind the scenes and pulls the tabular results back as a chat response. And just as simply, the employee can ask the digital employee to graph that data and the digital employee will convert it to a graph.
  • Automation skills: There are a plethora of automation technologies that are very good at what they do, whether its orchestrating workflows, modeling business decisions, processing documents or even automating employee tasks using robotic process automation (RPA). The digital employee doesn’t seek to reinvent these automations; instead, it aims to leverage them and make them accessible to employees. Imagine an army of RPA bots that are now accessible to your employees when and where they need them through natural language. 
  • Integration skills: Employees spend a good share of their time working inside of business applications, such as Salesforce, Workday, etc. This work takes them away from higher value tasks like having conversations with customers/employees or solving problems and making decisions. The digital employee is their sidekick that can take on the more tedious task of updating records in those business applications through integration skills. Integration skills leverage existing APIs to let the digital employee perform work. Using its unique credentials, it will only perform work and access data with which it is allowed to work. 

Bringing it all together

To summarize, a digital employee, in IBM’s view, is the coming together of a complex set of technologies. Some already exist, such as natural language understanding, an integrated automation platform and an integration platform. Some are groundbreaking innovations, such as Verdi, with its built-in planning and understanding capabilities. These are brought together to empower businesses to compose digital employees to augment any part of their human workforce.  

The example we discussed here was a human resources use case, but watsonx Orchestrate is already being envisioned to augment sales teams to automate work in Salesforce, in insurance to help speed up the claims approval process and in finance to help relationship managers be better prepared for customer meetings. Digital employees help automate focused use cases for employees. Build the right skills and digital employees can automate any line of business. Fortunately for watsonx Orchestrate, building skills is a low-code exercise that a line of business can do for themselves.

Debunking common myths

  • Digital employees will replace the human workforce: Digital employees — and AI, in a broader sense — are just the latest technology enablers. Just like when computers entered the workforce, computers didn’t replace workers, they elevated the employees to be more productive. For example, before a cash machine, people had to handwrite receipts and hand calculate totals. Letting a machine take care of that for them didn’t replace them — it enabled them to do other, more meaningful work. A digital employee aims to do the same for today’s employees. Take on the mundane or repetitive work to increase employee productivity, employee impact and job satisfaction.
  • Digital employees are the same as conversational bots: Digital employees focus on automation; it’s about building automations and making them available to your employees so they can use them on the fly to get work done. Conversational technology, on the other hand, focuses on the dialog and requires that you create a static dialog tree that limits how the employee can interact with the solution. With the AI in digital employees, there is no pre-determined dialog or workflow. The dialog is enabled on the fly by enabling skills. The more skills you add, the more things you can tell the digital employee to do. Additionally, conversational bots tend to be agnostic of the user. They can execute one-off stateless requests, whereas digital employees maintain a per-employee context and are able to provide a stateful, personalized experience. Lastly, digital employees are designed to be rolled out in multitudes. You could have a dozen HiRos added to an organization, each unique to the team its working with. For example, the North America HiRo uniquely equipped to work with the systems and data specific to its geography, as compared to the EMEA HiRo and the Latin America HiRo.
  • RPA already automates employee tasks: This is true, and digital employees don’t aim to replace RPA bots. They aim to make them accessible to employees at the time of need. Furthermore, digital employees allow employees to mix and match RPA bots, automations and integrations to work together on-the-fly.
  • Business process and case management already orchestrate workflows: BPM and case management systems fulfill the need for end-to-end process workflow automation and provide the regulatory controls needed by enterprises for the overall business processes. Digital employees don’t aim to replace these workflows, but rather aim to fill the gaps that these workflows cannot address in a scalable way. The small tasks and the time spent accessing and working with multiple systems that consume employees daily — this is the niche digital employees aim to fill.

Learn more

Digital employees are the way of the future. We know it. When and how is in our hands. It’s on us to shape that future responsibly, in a way that advances organizations to elevate their employees to new levels of performance and innovation.

watsonx Orchestrate is IBM’s point of view on the future of digital labor. Come learn more about it and experience watsonx Orchestrate firsthand.

Also, make sure you check out The Art of Automation podcast, especially Episode 2 in which Jerry discusses the topic of Digital Workers with Rania Kalaf, co-inventor of the Verdi technology mentioned in this chapter:

The Art of Automation: Landing Page

Was this article helpful?

More from Automation

5 SLA metrics you should be monitoring

7 min read - In business and beyond, communication is king. Successful service level agreements (SLAs) operate on this principle, laying the foundation for successful provider-customer relationships. A service level agreement (SLA) is a key component of technology vendor contracts that describes the terms of service between a service provider and a customer. SLAs describe the level of performance to be expected, how performance will be measured and repercussions if levels are not met. SLAs make sure that all stakeholders understand the service agreement…

Scale enterprise gen AI for code generation with IBM Granite code models, available as NVIDIA NIM inference microservices

3 min read - Many enterprises today are moving from generative AI (gen AI) experimentation to production, deployment and scaling. Code generation and modernization are now among the top enterprise use cases that offer a clear path to value creation, cost reduction and return on investment (ROI). IBM® Granite™ is a family of enterprise-grade models developed by IBM Research® with rigorous data governance and regulatory compliance. Granite currently supports multilingual language and code modalities. And as of the NVIDIA AI Summit in Taiwan this…

Maximizing SaaS application analytics value with AI

5 min read - Software as a service (SaaS) applications have become a boon for enterprises looking to maximize network agility while minimizing costs. They offer app developers on-demand scalability and faster time-to-benefit for new features and software updates.  SaaS takes advantage of cloud computing infrastructure and economies of scale to provide clients a more streamlined approach to adopting, using and paying for software. However, SaaS architectures can easily overwhelm DevOps teams with data aggregation, sorting and analysis tasks. Given the volume of SaaS…

IBM Newsletters

Get our newsletters and topic updates that deliver the latest thought leadership and insights on emerging trends.
Subscribe now More newsletters