The Company

Apromore is an open source process mining vendor co-founded in 2019 by Marlon Dumas, Marcello La Rosa, and Simon Raboczi. The company is a spin-off from research undertaken at both the University of Melbourne and the University of Tartu (Estonia). Headquartered in Melbourne, Australia, the firm initially raised around US$8 million in seed/research funding, and that was followed by a series A round of US$5.3 million in July 2020. The company has approximately
35 employees and is recruiting more.

Though Apromore has only been around for a few years, the product and team behind it have a rich history going back to 2009. As befits a product developed in academia, it is the result of 10 Ph.D. theses and over 150 scientific publications. As it is open source, its underpinnings are well known and have been widely shared in the process automation sector over the years.

The Technology

Put simply, Apromore provides a commercial (Enterprise Edition) process mining application built on top of its own open source platform. This platform consists of an architecture running graph mining and AI/ML algorithms developed for key process mining activities. Apromore believes their analytical capabilities are more precise than the competition’s – a bold but likely very defensible claim for at least some of the activities since their algorithms are published, widely used, and freely available. But at Deep Analysis, we are looking for innovation and differentiators, not comparisons. Apromore struck us because, though it labels itself a process mining vendor, it is more than that and could accurately be called a performance checking/optimization/monitoring system.

Let’s dig a little deeper, starting from the top.

The UI is designed around the concept of a Google Drive, with everything from modeling to process intelligence in one place, and functionality and views accessed through files and folders. You can build your folders here – for example, one per project – and essentially build out a case-style process management system. We found the UI to be good and well thought through, and it was easier to move from one process view to another than in some systems we have seen.

The functionality exposed in the UI is expansive, ranging from simulations, to the mining activities themselves, to the ability to check conformance or, for that matter, undertake some variant analysis. Underlying this is the “Split Miner” algorithm designed to accurately and quickly reverse-engineer business process mining notation (BPMN) models from captured log data (see Figure 1). This algorithm is supported by a conformance-checking function that Apromore calls Behavioral Alignment, which differs from traditional methods to check that behaviors in the process log align with the model, and vice versa. The difference is that Apromore can check alignment against the entire process at once, versus traditional methods that align one trace at a time. This is technical stuff indeed, so suffice it to say that, in our analysis, Apromore is a highly advanced architecture and set of services that can run both in the cloud and on-premises.

Apromore has all the bells and whistles of an advanced process mining suite like Celonis or Signavio, yet two things stood out for us. First, Apromore is fully integrated (as of Oct. 21, 2021) with the task mining product UltimateSuite. Rather than bridging the gap between task and process mining, this integration brings them together in one place, potentially a bit of a game-changer. This leads to the second standout item: the company’s roadmap. Typically, we stay away from vendor roadmaps as they can be a product of guesswork, and to be clear, Apromore will have to deliver on its roadmap promises. That said, while the Apromore product today is focused on the “as is” and the “what if” scenarios, as are their competitors, Apromore has plans to release, in 2022, robotic process mining (in collaboration with UltimateSuite), predictive monitoring, and causal mining.

• Robotic process mining, which discovers digital routines in live environments to generate RPA scripts, will be released in Q2 next year.
• Prescriptive process monitoring, which delivers real-time action recommendations from predictive models, will be released in Q3 2022. All this work is based on research publications from the co-founders.
• There are longer-term plans to bring about causal process mining; in other words, observing causal effects on process outcomes and in turn making recommendations to address those effects.

Most process and task mining tools provide much-needed dashboards and visualizations of otherwise unfathomable log data and task minutiae. Apromore wants to go far beyond that and provide AI-driven process optimization tools that automate a lot of the analysis and provide actionable insights and recommendations as well.

In summary, the move by Apromore to combine the data from task and process mining into one data set, supported by its strong underlying AI analytics, offers potential for more automated process delivery in the future. That may come through automated RPA scripts or through generating accurate training data sets for AI-driven process management systems.