Marlon Dumas, co-founder and chief product officer of Apromore, delivered his company’s perspective for the Acceleration Economy Course, Process Mining in the GenAI Era. Dumas touched on digital, predictive process monitoring, artificial intelligence (AI) and generative AI (GenAI), and future process mining market directions. Below are highlights of our discussion.
Highlights
Company History (00:57)
Apromore was “born” in 2009 in the form of an open-source tool for process optimization. For about 10 years, it existed as an open-source project with contributions from different universities, research labs, and individuals around the world. There were roughly 200 or more open-source customers globally, spanning healthcare, government, and a long tail of company types including telcos, utilities, energy, and more. The company needed to scale up from open source so an enterprise edition was created with process mining, process modeling, and simulation.
Ask Cloud Wars AI Agent about this analysis
Well-Rounded Approach (05:17)
Apromore’s years of experience taught the importance of combining different techniques together; for example, combining process mining with Six Sigma (there were Six Sigma users of the open-source platform); how to tackle compliance issues which are prominent in the government sector. The company also learned the importance of combining process documentation and process modeling with process mining and simulation. “It taught us there is a lot of diversity in the field and that you have to tackle problems from different perspectives and that is something that we bring to our customers,” making their process-mining initiatives well-rounded.
GenAI in the Company’s Product Vision (07:46)
A lot of the focus of process mining has been on the “discovery” phase — understanding a process, discovering the process map from the data, discovering the backbone of the processes, the exceptions, and navigating through that “spaghetti.” Simulation allows customers to do what-if analysis and test different scenarios. But there’s a big gap between identifying a friction point in a process and identifying possible improvement actions. “Our bet is that generative AI will allow us to have a much more structured and comprehensive approach to redesigning the process,” Dumas says.
GenAI Deliverables (11:21)
Apromore has its GenAI copilot in beta mode. The technology in large language models (LLMs) is evolving rapidly; many more evolutions will happen in the next 6 months — they will not only change the quality of outputs but also change the capabilities that it provide to you. So it will stay in experimental mode until the second half of the year then go into general availability. Apromore is working with innovation departments in banks, telcos, and insurance, validating the hypothesis that LLMs allow customers to take findings from a process mining “agent” and propose sensible improvement options.
Digital Twins to Optimize Operations (18:57)
Apromore has been surprised by the level of adoption of digital twins. While the company thought it would take five years to convince clients to trust digital twins, some of its Fortune 500 customers are now using digital twins for business processes with 300 to 500 activities. To generate insights at a detailed level, the digital representation should closely mirror the intricacies of the real-world operation. For example, in a large insurance company, there are a wide range of claim amounts, from the tiny claims to very big commercial liability claims.
Predictive Process Monitoring (25:47)
Another differentiator for Apromore is root cause analysis specifically, the capability to efficiently analyze complex sets of features. This enables it to extract specific characteristics inherent in the process and pinpoint the particular types of cases that cause the greatest pain for companies in terms of a given KPI. The other area is predictive process monitoring, which Apromore began offering in 2017.
Accelerating Time to Value (32:34)
Time to value means shortening every bit of the trip from the moment you have reasonably refined data, to the moment you get your insight, to the moment you start implementing. Apromore’s low-code/no-code approach enables business teams to do the whole journey from data to insight without having to go back and forth with IT. What could be weeks becomes two days.
Strategic Project Scoping (35:16)
Customers need to narrowly define the scope of what they want to achieve, perhaps from some top-level KPIs that are hurting the business and from some core processes. And that scoping, in Apromore’s experience, requires you to think about two things: time to value and time for value. Your time has to translate into value. So, customers need to scope problems to focus on core processes and to select very carefully which processes, and which variants, to optimize.
Core Use Cases (37:16)
The banking, financial service, and insurance industries have been strong adaopters of process mining. Customer service is an area that has high adoption; service requires optimizing in different dimensions with respect to time, costs, and other factors that drive operating margin and optimize every touchpoint. Compliance is a back-end process that benefits from process mining.
2024 Predictions (41:47)
There is increasing demand from operational managers to be able to access the process mining tool. “In two years time, the traditional line between process mining as a tactical optimization tool and more operational intelligence dashboarding tools for operational optimizations are going to merge and we’re going to have a convergence of operational intelligence with process mining colliding together,” Dumas predicts.
Apromore Product Demo (45:30)
Dumas walks through a demonstration of Apromore’s offerings, starting with its digital twin capabilities and making a connection with traditional process discovery. Apromore enables its customers to take advantage of streamlined models, with data displaying the benefits of having a better, customized process. Two key features of the platform are root cause analysis and predictive process monitoring, which both allow for deeper exploration of processes.
