IBM Automation Insider

A bimonthly round-up of useful information to help you automate all types of work at scale

November/December 2019

5 articles

31 min

Why you can’t be cognitive without automation

By Katie Sotheran and Tuck Satterfield

Cognition is the process of collecting, remembering and analyzing inputs to derive insights and understanding. Humans have been doing it from our beginnings, and we’re good at it. So good, we’ve built tools to facilitate and automate cognitive tasks, helping us to think and learn more deeply or faster.

Now, cognitive technologies like artificial intelligence (AI) and machine learning (ML), along with the computers and devices running them, are part of everyday life — in our cars, phones, homes and offices. They underpin any number of daily digital transactions. According to Pew Research Center¹, 46 percent of Americans use voice assistants, such as Siri or Alexa, which use natural language processing and some ML to understand and respond to our requests. No big news here: humans are pretty comfortable interacting with available AI or cognitive technologies.

The bigger news is the shift in how these technologies are used, or plan to be used, in enterprises. AI is no longer about making people more productive. It’s about making whole workflows and systems more productive. As these systemic transformations become the norm, simple time savings will become a smaller part of the overall goal. Broader measures that reflect the health and quality of entire work processes and client journeys, like Net Promoter Score (NPS), will become the aim.

Achieving cognition at a scale to meet these new and further-reaching business goals will require business process automation. Automation is the mechanism for how AI and ML get translated into action. Automation is how you put AI to work.

Why you can automate without being cognitive, but you can’t be cognitive without automating

Automation has been around nearly as long as cognition, but when we talk about cognitive enterprises we’re talking about a more intelligent version of automation. Cognitive enterprises use software-based labor to execute entire workflows — rather than just certain tasks — which can fundamentally reshape how work gets done. These new intelligent workflows run digitally, in machine-time, with vast amounts of data flowing through them at a pace and volume humans simply can’t handle.

Think about all the data in your workflows and how it gets addressed by ML or AI to create decisions. The AI that handles the data needs automation capabilities — such as robotic process automation, workflow, or business rules — to execute those decisions. In other words, AI without automation is like a car with self-driving intelligence but no engine: you have the smarts but not the mechanism to execute on it.

Conversely, you can automate without being cognitive. Screen scraping or task automation can be done at scale with significant benefits. But these are bolt-ons that can provide some velocity and efficiency advantages; however, they don’t shift your business model to take advantage of the cognitive era or truly transform how work gets done. Keeping with the same smart car analogy, automation alone might give you a faster car but, without the AI, you still need to figure out how to get from point A to B and drive there yourself.

In sum, business continues to evolve. Cognitive and automation technologies are evolving even faster — in ways that are complementary. Gartner recently used the term “hyperautomation” to refer to “the combination of multiple machine learning (ML), packaged software, and automation tools to deliver work.”² Whether you call it hyperautomation or intelligent automation or something similar, it’s about making entire work processes and systems more productive. This requires a range of cognitive and automation technologies coming together to deliver both the intelligence and the power to put the intelligence into action.

If you’re interested in learning more about what it means to be a cognitive enterprise and how IBM is walking its talk:

  • Listen to the 19-minute interview with IBM Automation executive, Gene Chao, hosted by Daniel Newman, Principal Analyst at Futurum Research.
  • Read the seven keys to success.


  1. “Nearly Half of Americans Use Digital Voice Assistants, Mostly on Their Smartphones,” Pew  Research Center.

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.

How TTI is using RPA to remix work and spend more time with customers

By Cheryl Wilson

 “ … we’ve put 10 bots into production that will save 34,000 hours of staff time this year.”

– Jake Brown, vice president of IS Global Development at TTI


TTI, Inc. — a major global provider of electronic components and products — had a problem common to many companies across industries: skilled employees were spending too much time on lower impact tasks, leaving less time for higher impact work. In one instance, TTI employees were tied up for hours managing and tracking electronic components — time that could’ve been spent building relationships with customers … your classic knowledge worker’s dilemma.

While inventory management is important for TTI, especially with 800,000 units in stock in North America alone, the related processes were slow and manual, involving many departments. Just handling new pricing models from suppliers took hours of crunching numbers and manipulating spreadsheets. And the only person who knew the process — all 26 steps of it — was retiring.

“We had inefficiencies which could lead to delays and downtime,” Brown said. “And this affects our customers, suppliers — all of our relationships.”

How TTI solved its knowledge worker’s dilemma

After seeing what another company achieved with IBM, TTI scheduled a Discovery Workshop. During the workshop, they built a bot to handle the new pricing model process using IBM Robotic Process Automation with Automation Anywhere, with help from IBM Business Partner BP3 Global.

“They really proved this platform could handle something that complex,” Brown says.

The new pricing model process can now be executed in a fraction of the time with almost no oversight, freeing the employee to work on higher value tasks. TTI also retains the process knowledge when the staff member retires.

But this performance story goes beyond inventory management. The use of IBM RPA with Automation Anywhere quickly spread as other departments saw the value of more efficiency, cost savings and time to focus on more important work.

In billing: TTI’s customers need to know which items incur tariffs. Before automation, it took five finance employees almost two weeks every month to gather and deliver this information. Using the IBM tool, a bot was written in two days that could handle all 400 hours of manual effort, allowing staff to work on other things.

In quality control: Recalls used to take weeks and involved a lot of routine research. Now, with the click of a button, a bot does it in about four hours. Employees use the time saved to talk with the suppliers so they’re prepared — something bots can’t do.

In IT: Staff gained productive hours by using the tool to write bots to perform scrubbing, saving them from writing scripts all day. 

Back in finance: Before an employee can reach out to a customer about a disputed invoice, they need to pull data, which previously took days. Now, a bot tracks down the historical data so employees can have meaningful conversations with customers more quickly. Icing on the cake: automating the dispute research saved USD 600,000.

What’s next?

“It sounds like we can cut back on people, but that’s not it at all,” says Brown.

Allowing staff to focus on important human relationships is critical to TTI’s philosophy and success. Using RPA is just the first step in remixing work — blending automated and human processes — to drive growth and free people to work on things that can’t be automated.

If you’d like to learn more about TTI’s story, watch Jake Brown — in his own words.

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.

Eight tips to drive better customer experience from your process models

By Jeff Goodhue

We know this: if you don’t provide the best customer experience, your customers go elsewhere. One way to get them to stick around: speed up the time it takes to complete core operational processes by automating areas with high-cycle time, low-value-add or manual decision points.

You’ve taken the right first step towards achieving this if you modeled the processes. When I ask clients for their as-is process models, more often than not they say one of the following:

  • "Oh, we never created one of those.”
  • "We have them, but they’re outdated.”
  • "We have one. Let me get it.” (This is my favorite, especially when they find at least three versions!)

Your as-is process model is as important as the red "You are here" dot on a map. Spend sufficient time on these models to understand current problems and to ensure you don’t miss process improvement opportunities.

Once you have your as-is process model, you’re ready to create a to-be model (Figure 1) by identifying desired improvements and optimizations through process analysis.

screenshot of a process diagram

 Figure 1. To-be process being analyzed for cycle time in gray against problems in red

After you know where you are and where you want to go, the next step is to make those models work for you by automating them to reduce cycle time, improve customer experience and free your team for higher value work.

Eight tips to apply business automation to your process models

Note: These tips are especially useful for business analysts and process improvement leaders. Along the way, we refer to several IBM business automation tools for practitioners who want to get started right away or just try something.

Tip 1: Execute your process

First take an inventory of your process models and consider the cost and value of business automation through execution. For example, how much would your knowledge workers and customers benefit from integrated data retrieval, automated routing, real-time visibility, audit trails and other process automations?

Once you decide the right process to execute based on the anticipated business value, there are processing mapping tools, like IBM Blueworks Live, that can help you prepare your model for execution even if you didn’t model it with execution in mind.

Watch this demo on low-code process modeling and execution to see this in action.

Now that you’re executing your process, the next seven tips focus on using intelligent automation services to gain even more value — to further reduce cycle time, improve customer experience and free your team for higher value work.

Tip 2: Replace process branches and gateways with knowledge worker-managed decision services

As you analyze your processes, look for the following characteristics in branches and gateways, such as those shown in Figure 2 below: high frequency, significant volumes of data required, flexibility of change benefits, or regulatory and business audit requirements.

The business rules most likely present in these circumstances — either documented or in people’s heads — are good candidates to manage externally. By managing them outside the process, you enforce consistent business policy while empowering your business teams to easily simulate and update rules. In a faster moving world, the reduction in cycle time through decision automation directly improves customer experience and retention.

flow chart for access control

Figure 2. Process branches and gateways may represent reusable decisions

Try it

Log in to IBM Blueworks Live, and try the following steps:

1. Locate a process similar to Figure 2 above (or create your own three-activity process).

2. Right click the Determine Employee Access task and select Type -> Decision Task.

3. Click, drag and drop the Determine Employee Access task to the BPM System swim lane. You should have a process similar to the one in Figure 3 below.

flow chart showing access task in bpm system swim lane for automation

Figure 3. Decision task moved to the BPM System swim lane for automation

4. Double click the Determine Employee Access task, select the new Decision tab and create a new decision.

5. In the resulting decision diagram, double click the name of your decision and see the new details you can add plus the Decision tab where you can create your own rule tables.

6. Go back to your original process diagram — notice that the linkage to your new decision is retained for traceability and reusability.

Tip 3: Remove data entry errors with intelligent content extraction and robotic automation

Time-intensive data entry and document look-up tasks as shown in Figure 4 can be automated with a combination of intelligent content extraction and robotic process automation to significantly reduce human error and to save time.

screenshot of error message

Figure 4. Process analysis makes it simple to find high-cycle times and data entry problems

Digital and scanned documents can be sent to a content analysis tool and automatically classified as a specific document type. Then, relevant data is extracted in a structured form to be used by downstream processes. If the original task also includes user interfaces that are difficult to automate with APIs (for example, data entry) or aren’t accessible behind secure interfaces, a robot can help automate the remaining entry with no change to the existing user interfaces.

Tip 4: Replace “string of pearls” patterns with APIs

When you see a series of user or system tasks in the same swim lane (a “string of pearls”) as shown in Figure 5, it may be an opportunity to combine and use APIs so you can remove repetitive tasks and significantly speed up process cycle time.

screenshot of series of tasks

 Figure 5. String of pearls process pattern leads to a good place for API integration

Try it

Log in to Blueworks Live, and try the following steps:

1. Locate a process similar to Figure 5 above (or create your own string of pearls process).

2. Press Ctrl (Command for Mac OS) and click to select the three activities in a row (you can also Shift select them by clicking the first and last).

3. Right click on one of the three and select Convert to Subprocess.

The result is a subprocess activity (that can be collapsed and expanded), which contains the original three activities. Figure 6 contains an example.

screenshot of a subprocess activity with three activities

 Figure 6. Three activities converted to a subprocess

To move forward, consider APIs that may already exist or search your organization’s API development portal. If no API is available, look into creating a new API using a tool like IBM Cloud Pak for Integration or continue reading this article for more options.

Tip 5: Use robots for nonvalue-added tasks with high-cycle time and legacy systems

During process analysis, when you see high-cycle times combined with no Value Add to the customer (as shown in Figure 7), it’s a good indication your knowledge workers may be spending time on tasks that don't merit their attention. You can also trace where your process uses systems that can’t be upgraded or reached with an API.

For those tasks that focus on user interaction, you can offer your knowledge workers a robotic assistant to help them with repetitive work. They can even control when the assistant will run and on which tasks.

screenshot of no value add message

Figure 7. Process analysis can tell you at a glance when cycle time is high and a task is not adding value

Tip 6: Infuse AI into value-added tasks with large volumes of data

Artificial intelligence (AI) provides the most benefit in value-added process tasks that improve customer experience and require access to large volumes of data (for example, recommendations and situation detection). Think about having to answer a customer question without having time to do the research in order to make the best recommendation. High-cycle times can indicate that knowledge workers are spending a long time finding and combing data to determine the next-best action for their customer (Figure 8).

screenshot of value add message

Figure 8. Process analysis can tell you at a glance when a task adds value to your customer

Recommendation: Put AI to work indexing your organization’s structured and unstructured data sources, such as document repositories and file systems. Then add an AI service, such as IBM Watson Discovery, to your process to help knowledge workers retrieve and easily visualize the information they need.

Tip 7: Launch your processes with virtual assistants

How do you launch your enterprise processes? Do you click in a portal or send an automated event, message or API call? You can directly drive work to the right process and team — seamlessly passing context and delivering improved customer experience — by integrating your processes with virtual assistants, such as chatbots.

Most virtual assistant projects start standalone with small connections to data sources for Q&A patterns such as FAQs.

Try it

See a chatbot in action using IBM Watson Assistant:

1. Go to the IBM Watson Assistant page and click the Get Started button.

2. Sign up for a free IBM Cloud™ account as required.

3. Once you log in, you can create your first assistant.

4. If you would like a step-by-step tutorial, try this.

As assistants become more intelligent, they move from conversational to actionable. They use cloud functions and APIs to improve integration from front-end chatbot to back-end operational processes. Consider which processes would benefit from a conversational interface to gather required information and automatically launch. The resulting pattern would look like Figure 9 below with multiple start options for added flexibility.

screenshot of multiple start options

Figure 9. Multiple process start types provide flexibility and allow the virtual assistant direct access

Tip 8: Combine rules with machine learning for the best recommendations

Do your decision tasks — identified in the second tip — require more data types than you can process with rules alone? Do you want to use the predictive scoring models from your data science team in your operational processes?

Begin with a unified decision model that uses input from traditional structured sources (like a customer or an account) and new machine-learning-based predictions (such as likelihood to accept product or risk score). Ensure the business team is involved in the initial decision modeling and business rule design to maximize reusability of the decision services. 

Remember these decisions can be adjusted using both business rule policy updates and predictive model parameters. Also, advanced patterns such as rule-driven predictive model selection allow your team’s decision models to prescribe the next-best action for your business.

Watch this demo to see an example of infusing business rules with machine learning.

Pulling it all together

If you need to reduce cycle time, improve customer experience and free your team for higher value work, it all starts with a solid process model. After that, analyze the model to identify opportunities and apply business automation using the eight tips above. Don’t forget to look for advanced combinations that can amplify business value, such as decisions and machine learning or intelligent content extraction and robotic automation.

Figure 10 models the tips outlined in this article using Blueworks Live. You can use this to follow the path from process modeling to business automation.

screenshot of automation model

Figure 10. Path from process modeling to business automation

Finally, if you’re not sure which process to start with, consider a free one-day automation workshop. Or if you’re ready to scope your MVP, get started with a four-day Garage session.

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.

“How can I inspire my employees to embrace automation and AI?”

By Benjamin Chance

Technologies change, use cases evolve, markets are disrupted, but it’s employees who remain the critical constant, underpinning a company’s ability to innovate and create better customer outcomes. As you move from simple task automation through process and workflow automation, the more engaged your employees, the more successful your automation transformation.

But changing to new ways of working with automation and AI — or intelligent automation — can be a complex enterprise conversation. You can find yourself fielding multiple questions or facing a spectrum of emotions, biases and narratives — from job loss to talent gaps. How do you create a virtuous cycle of automation-based transformation that embraces a diversity of thinking and workforce mixes, and doesn’t speed-bump your automation roadmap?

Proven practices for employee engagement

Through hundreds of client automation transformations, we’ve seen what tends to work and what doesn’t when it comes to creating a culture of employee engagement around intelligent automation.

The following five best practices can inspire employees to accept and even embrace intelligent automation, but they only work when there’s a clear change strategy and when everyone’s speaking the same “automation” language (for example, what do we mean when we say “digital worker”?).

1. Include employees in workflow re-imagination — from the beginning.

The perfect place to start the hybrid — humans+digital — workforce conversation is when you’re imagining a to-be automated workflow; ideally, using design thinking. Your employees are your process experts. They know their data, customers, and ecosystem of stakeholders and adjacent processes. Let them imagine how they can work more efficiently with automation. Listen to where they feel automation is most needed or valuable. And discuss the associated risks and opportunities for all employees.

2. Involve employees in design and development.

Using agile principles, employees should be engaged in the design and development of the automation components that comprise an intelligent workflow so they can fully understand the interactions required in the hybrid workforce. If the right skills and opportunities exist, encourage employees to get hands-on, to experience and master new ways of working.

3. Inspire teams with new opportunities to grow their skills. Creating a continuous learning enterprise with structured talent roadmaps and training curriculums can motivate employees to develop new marketable skills and pursue the careers that best fit them and the company.

4. Get employees to participate in workflow deployment and continuous improvement.

Employees can learn the power, and limitations, of AI-driven processes by participating in the delivery and continuous improvement of the workflow. This ongoing interaction with intelligent workflows can create evangelists who provide “word of mouth” testimonials to colleagues.

5. Use AI capabilities to build personalized education journeys.

Scaling any transformation requires targeted, individual training roadmaps. Using employee analytics, AI and chatbots, you can provide better learning engagement and outcomes while demonstrating how you’re “drinking your own automation champagne.”

The best people to engage employees

Although employee engagement is inherently an internal initiative, kickstarting and scaling it across the enterprise may require outside help. Either way, it’s important to understand the key internal stakeholders best situated to successfully lead the engagement conversation.

Executives: Our most successful clients have started with a clear understanding of their change capabilities and roadmap from the top down. Having key executive stakeholders (for example, CHRO, CEO, CFO) collaborate to lead a conversation around change across all levels of the enterprise sets the foundation for positive employee engagement. 

Automation center of excellence (CoE): An automation CoE can be, and usually is, your central source for clear messaging and communications concerning any automation transformation. The CoE should have a remit to handle and shape communications as well as enable business unit and IT-specific communications.

Process and functional owners:  Make sure team leaders understand the potential of automation to move their teams to higher-impact activities through intelligent workflows so they can develop and clearly communicate plans and opportunities.

In summary, inspiring your employees to embrace automation and AI is a critical success factor. Employees are the past, present and future of our businesses. When we can align the right tools with their aspirations and motivations — while decreasing any stress or discomfort — we create space for them to contribute exponential business value in new and existing markets.

To learn more on this topic, download the Forbes Insight report Intelligent Automation: How AI and Automation are changing how work gets done (PDF, 2.3 MB).

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.

A human-centered automation tool for building intelligent digital workers

By John Davies

We’ve been talking about the need to solve the knowledge worker’s dilemma for much of this year.

The story goes like this: you hired Lisa, a loan officer, for her high-impact skills, such as industry expertise, client relationship building and finding creative ways to grow the business. But she spends a lot of time on lower-impact work like inputting data, managing documents and creating reports. This reduces the total time she can spend adding more value to your business, such as increasing the number of loan applications per month.

In a world where lower-impact work increasingly swamps talented workers — where the rate and scale of digitization creates information faster than workers can consume and act on it — it’s no surprise tech vendors are introducing solutions to help grow productivity and improve outcomes with the help of digital workers.

IBM Automation recently announced our own digital worker capability with the mission to bring digital scale to knowledge work. For many companies, this means enabling a small number of employees to create great customer experiences at scale, to produce better products and services. For others, it’s all about reducing costs, increasing agility and flexibility, or improving compliance.

Whatever your business goals, we explain the new capability and how it works.

A human-centered tool for helping knowledge workers

By bringing together automation and AI, you can now use the modular IBM Cloud Pak™ for Automation to build digital workers — digital workers that not only automate lower-impact work like data entry and data extraction, but also assist knowledge workers in higher-impact work like task prioritization and decision making.

The tool was designed with the human knowledge worker at its center. Instead of being aligned to a specific task (like a bot) or process (which crosses multiple functions), the IBM Automation digital worker is aligned to a real job role, such as mortgage underwriter, data entry clerk, customer support assistant, HR recruiter or sales assistant. This more intimate alignment to day-to-day work makes it easier for knowledge workers to offload work where they don’t add value.

In three minutes: How the new digital worker tool works

With IBM Cloud Pak for Automation, business users can easily create digital co-workers to execute specific parts of their jobs. This is how it works — at a high level:

1. Model the job role

A built-in assessment capability makes it easy to model existing human job roles: the sequence of tasks performed, the time spent performing those tasks and any required interactions with other roles. The assessment helps you prioritize where to deploy digital workers and set baseline key performance indicators (KPIs) so you can determine productivity gains.

2. Build a digital worker

To perform work within a specific role, you build digital workers by teaching them skills, ranging from simple to complex. You select these skills from a pre-built catalog. Custom skills can also be created. And you can teach digital workers more skills after they’re deployed to further assist their human counterparts.

The digital worker skills line up under the following essential competencies:

“Understanding information” skills:

  • Data extraction from docs
  • Document classification
  • Natural language classifier
  • Tone analysis
  • Language translation
  • Visual recognition

“Making decisions” skills:

  • Rule-based decisions
  • Machine learning (ML)-based predictions

“Taking action” skills:

  • Read an email
  • Send an email
  • Take actions in Excel
  • Start a workflow
  • Run a robotic process automation (RPA) bot

3. Set business controls

When you delegate work to a digital worker, you need to be able to trust the work will be done right. Using the tool’s business controls, business users can easily manage the digital workers they deploy. It’s not much different from managing human workers.

Digital workers perform within guardrails based on their experience levels. Think of guardrails as parameters or confidence levels, set by the business, within which digital workers can operate autonomously. For example, a loan processor digital worker can be assigned to only process loans under USD 5,000. After the digital worker proves itself, the human supervisor can easily update the guardrail to give it more responsibility, such as increasing the threshold to USD 10,000 (Figure 1).

screenshot of guardrails for loan processor

Figure 1. Business users set and adjust guardrails to control the scope of digital workers

The tool also includes built-in transparency to ensure compliance and auditing, and built-in monitoring of digital workers so adjustments can be made based on the quality and quantity of work performed. There’s a built-in management “cockpit” that provides one consolidated dashboard for work done by humans and digital workers.

Net net: Digital workers bring together a variety of automation technologies — RPA, content management, data capture, AI and more — to do work on behalf of human workers in real job roles. For instance, if you want your new account specialist digital worker to pull information from a new application and input it into your customer database, you could use the “run an RPA bot” skill to transfer the information. If you want it to also decide whether to accept or reject the application, you can add a rules-based decision skill to make this determination.

Solving the knowledge worker’s dilemma requires focus on the knowledge worker, and not on automating discrete, routine tasks. For this reason, digital workers can be a smarter approach to augmenting knowledge workers because they:

  • Use job roles as the framework to align the work that digital and human workers do.
  • Give control of digital workers to the knowledge workers themselves through interactive dashboards and adjustable guardrails.
  • Automate more use cases by using a full set of AI-powered automation technologies.
  • Make it easy for humans and digital workers to interact using workflows, emails, shared documents and collaboration tools.

If you’re interested in really experiencing how digital workers could work for you, think about participating in an IBM Automation Design Thinking Workshop. You’ll work side-by-side with IBM designers and automation experts to create first-of-a-kind MVP (minimum-viable product), experimenting with the new tool.

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.