There’s a lot of buzz—and buzzwords—around intelligent automation, none more so currently than process intelligence, a term often used interchangeably with process discovery, process mining, and task mining. To set the record straight, Benjamin Lingard of Reveal Group sat down with Steve Tsuchiyama, Vice President and General Manager, Process Discovery at Automation Anywhere, to answer the question: what are we talking about when we talk about process discovery and process intelligence?
Benjamin: Automation Anywhere has a very specific definition for process intelligence, correct?
Steve: Yes. We wanted to break through the noise, as the mixed terminology around process intelligence can be confusing. At the highest level, process intelligence is a component of process discovery, an AI-driven tool to continuously acquire process data. Just like business intelligence tools are used to collect business data.
How process intelligence can benefit your organization
Benjamin: So, it gives you an understanding of how your organization is running today?
Steve: Correct. Process discovery combines traditional task mining with process intelligence capabilities (computer vision and AI) to provide you with clear—and accurate—visibility into the current state of your processes. With process discovery, you get a continuous stream of process data, at scale, across any system or any application. Whether an ERP system, SaaS application, or productivity app, you get insight into how it is being used by people as they go about their daily work. Once you know how your business is running, you can drive initiatives such as process automation, digital transformation, or enterprise optimization.
Benjamin: Automation Anywhere has achieved good, solid growth in process intelligence—between 35 and 40% year-on-year growth. What are the reasons for this growth? What challenges are you helping organizations solve?
Steve: As companies pursue their automation initiatives, it's easy to identify the proverbial low-hanging fruit. After they maximize those opportunities, they try to identify more complex opportunities. But they’re probably talking to just one or two people involved in a process. Because they don’t understand all the variations a process can have—different people can do it differently—they get into a cycle of constant rework. After we demonstrate our Process Discovery solution and start to unpack these complex processes—not just the tail ends of them—they’re able to quickly automate them.
Benjamin: That brings up another question regarding organizational operating models. It's easy to slam some technology in and achieve early success. But as you try to scale, it becomes more difficult. Are you seeing organizations starting to put in place repeatable methodologies—that is, operating models—regarding process discovery?
Steve: Organizations typically wait for lines of business to bring them opportunities. If an opportunity meets a certain threshold of value, they automate it. But if they were more proactive—I’ll even use the word aggressive—to go out and look, they would uncover more opportunities. They could then weave this discovery into their operating models.
But I see many automation adopters that still don't have their operating models fully baked. To put process intelligence in, they need governance. That's what’s going to take this to the next stage of growth: establishing that governance model.
Benjamin: Shouldn’t they start with the problem statement or vision that the technology solves or supports, as opposed to starting with the technology?
Steve Yes, but we see a lot of technology-first approaches. A lot of companies collect massive amounts of data regarding users and how they work and how they interact with applications, down to how much time they take to perform certain actions. But these are just surface insights. You still need people to transform insights into action. That's what we're trying to do: to set up our own kind of operating model so that people are trained on how to use our Process Discovery solution and how to capture, interpret, and act upon those insights.
How process discovery and process intelligence affect your people
Benjamin: But with this, are we pushing people into data scientist roles, given the amount of data they must interpret?
Steve: It's still too early to tell whether our analysts need to be data scientists, too. We have a few customers where the data science team takes the raw data, looks for trends, and then partners with the process team to identify opportunities for improvement. But mostly we're seeing process or continuous-improvement analysts are the people who really know how to take advantage of process automation tools. And when we bring in partners like Reveal Group, they've helped accelerate that and set up governance models, returning early success.
Benjamin: Speaking of governance, are you seeing interest in process intelligence from chief risk officers, or do you expect to in the future?
Steve: One of our customers is an investment bank in New York, and they have something called “blue-sheet reporting” for compliance., where they must report every day to regulatory bodies on the trades they’re making. Before, it was a manual process, error-prone, and took forever. When they started using our Process Discovery solution, they were able to automate all the manual data entry and reduce errors. This is critical because once you fall out of compliance, that's a red flag to regulators. We’re exploring this governance opportunity in regulated industries more deeply in the coming months.
Benjamin: What tactical steps do you recommend organizations take when they start their process discovery journeys?
Steve: We tell people that you don't have to hit a home run right away. Just begin! Don't be afraid to experiment, otherwise, you’ll navel gaze for too long. If I can continue the baseball metaphor, single hits are okay when you get started—get your people on base to build some momentum. You don't have to uncover millions of dollars of opportunity right out of the dugout. Just get the bases loaded and then look for those big opportunities.
Benjamin: Can you talk about how organizations can justify the necessary spending on process discovery technology and consulting services?
Steve: If you look at it purely from the effort involved in manual process discovery today, that’s an efficiency gain right there. Although that doesn't immediately justify the entire investment, there are two other buckets we generally use, and if you fill each one, you get a much bigger number. This is how you build a business case. A second bucket is that you get throughput at scale from process intelligence because without the technology you’re only able to document the processes managed by one, maybe two, people at a time. Whereas with Process Discovery, you can broadly deploy across hundreds of users and capture everything that they're doing—not just one process—and analyze all of it.
Then, you have a third bucket where you can build a case on the automation side. Now, this is where things can get a little sticky because you can have instances where the automation team wants to take credit for the RPA aspect of it. But you can argue that those automations wouldn't have been possible if you didn't uncover the opportunities. That requires a little negotiation.
Benjamin: Can you explain how process discovery and process intelligence technology work, and how invasive they are?
Steve: Process Discovery needs to observe the workers for a set two-to-three-week period, but just once. It’s non-invasive. We deploy agents on desktops to record keystrokes for that one time. Once you have the data, you analyze it to surface opportunities. And you can reuse the data—no need to redo it—to look for more automation opportunities.
Benjamin: How closely should automation and process teams work together?
Steve: I was in London a few weeks ago, visiting a wealth-management firm, and they had both their process improvement and automation teams in the room together. And I asked, “Do you work together?” And the process team said, “Yes, we try to understand the process, look for opportunities to improve that process, then hand it to the automation team, and tell them, ‘Hey, now this is standardized so that you can go find opportunities.’” It’s fantastic to see the value from doing that. I'm hopeful that happens more frequently as the market evolves.
Benjamin: How do you see process discovery and process intelligence evolving, either in terms of the technology or the use cases?
Steve: I think today, there are still some technical limitations on how broadly we can deploy process intelligence across an organization. We need to make some improvements from a technical perspective. Then, there’s this concept of creating a work “system of record,” which is: how can people across teams stitch together all the functional areas and workflows that connect them? I don’t think the technology can handle that yet, but that's where it’s going. No one company will be able to do this, so we’re going to see a lot of partnering and overlap before we can give organizations that holistic end-to-end solution. Right now, we’re focusing on how we can integrate process discovery into RPA as seamlessly as possible.