Workflow lies at the heart of most business operations, from the smallest single process to the largest and most Byzantine corporate empires. Yet they are not a product, one never sees a single line item SKU on an invoice for `3 dozen assorted workflows’. But workflow as a concept can end up producing those `3 dozen assorted workflows’ if the right tools are available.
Building, or finding and pulling together the products that create those workflows has been the goal of Matt Calkins for some time. As the founder and CEO of Appian, his starting point was low-code applications building tools, But of late the company has gone on the acquisition trail as its route for expanding the end-to-end reach of integrated tools available to users.
Last year it extended one of those ends with the acquisition of Robotic Process Automation (RPA) specialist, Novayre Solutions SL, developer of the Jidoka RPA platform. This year, and just announced at the company’s recent European Conference, it has filled out the opposite end with the acquisition of Berlin-based Lana Labs and its LANA Process Mining applications. These help users discover the actual work patterns being used inside an organization by ploughing through system logs looking for common actions and sequences.
The objective here is pretty straight forward. By having LANA on-board as an integrated part of Appian, it will soon – as in, early next year – be possible for users to examine their current actual business processes as they work, compare that against a planned workflow, analyse its effectiveness plus identify weaknesses and improvements, and re-work the workflow so that it can then be implemented and operated automatically by RPA. In Calkins’ view they are increasingly seen as two halves of the same thing:
They're complementary. There's going to be a lot of benefit to having them both in the same platform. The process mining map and the workflow describe the same subject. It's a process. It's just the mining will tell you what's currently happening and workflows a set of instructions for the future. Our goal in this acquisition is to make these two products stronger by connecting them. We will discover new patterns with process mining, and generate workflows to automate and improve them. Mining analyses will become workflows automatically. After that the workflows will be sent back to mining to verify that the process is now running as intended.
Speaking during the conference keynote he noted that the process mining vendors, such as Lana Labs, lacked one important capability – the ability to take action on the findings their products unmasked. Making the join with Appian’s workflow and automation capabilities resolves this weakness in his opinion.
End-to-end now longer and more complex
Appian CTO Mike Beckley took this line of thought further, highlighting the importance of this capability as the scope of business workflows spread to encompass edge computing. This means that businesses will start to have many different systems, each potentially micro-optimising their area work, and we expect to be delegating tasks and activities to different systems that may be optimised for the edge or optimised for other workloads.
This is a world where distributed computing is becoming evermore distributed into smaller and smaller micro-services, even ones that are outside of your normal network purview with the edge. It becomes challenging for an enterprise architect or business to even understand where to start. Our vision is being able to not just design these enterprise architectures and automate your processes, but actually discover them in the first place.
“One of the reasons why we were so eager to do this acquisition, was we saw more and more clients coming to us with process mining tools, and saying, `now we have this backlog of processes to build, because we've got a 1,000 things we've mined and a 1,000 process maps. Let's get started.”
To that end, Beckley stated that Appian will continue to work with many of the other process mapping tools available on the market. The level of integration will not be as high as with LANA however.
This includes using Machine Learning with LANA and Appian to automatically sequence these events to visualise processes, and automatically visualise where there are problems in the processes. That the next step, for now, involves human intervention to manipulate the visualisations, do some data exploration, and ask some more questions. The Machine Learning is used to offer up the most likely visualisation for any given dataset.
Beckley often describes part of Appian’s role as providing guardrails to business analysis and management, and sees Lana Labs as an important step in lengthening and strengthening them. The goal is to automate the whole process of workflow optimisation and remediation, with AI playing a key part, but he acknowledges that goal is still some way off:
No-one’s at the point that you go from diagnosis to an automated solution or remediation of an issue. But at least now you can be set fairly sure that you actually know what you’ve got to fix and why you’ve got to fix it. This is an AI co-pilot that steps you through that, validates that you have fixed it at the end, and validates that you didn't create a new problem with a new bug in the system.
The acquisition also fits into the wider view Calkins holds about the goals of Appian. These revolve around three priorities: synergy, reliability, and low-code. The integration of technologies and tools, end-to-end, that synergy implies is perhaps the highest objective for the company now as it adds and extends the scope of what `end-to-end’ means for any business. When business processes and workflows pass beyond the firewalls of the internal network – which they do now for the vast majority of businesses - that integration has to be as bomb-proof as possible if they are to be managed at all, let alone well.
Giving orchestration good stuff to work with
To Calkins, `reliability’ means that everything has to be `enterprise grade’ in areas such as security and scalability, as well as the integration and the core technologies themselves. That, of course, includes low-code, which is where the company started. It now means much more than just reducing the workload of applications development, however. To Calkins it means factors such as: intuitive, graphical, empowering usability.
The scalability issue is perhaps the most important in at least one way with the Lana Labs acquisition. It traditionally refers to the ability of an application or service to scale across compute nodes, and more recently to an ever-lengthening end-to-end reach. It is possible to seen that this adds a further dimension, the scale of complexity across the lengthen of reach. Having the potential to analyse and map a process running across a cloud that stretches from the data center to `N’ sensors and `N’ monitors and `N’ manufacturing controllers, and back again, compare it to the planned workload, make the necessary adjustments robotically and have it all managed with increasing levels of AI-based automation could be the key component of business process orchestration.
With music, everyone can recognise a truly lousy orchestration, but it gets much harder when the orchestration is `OK’ but hard to work out how it could be better. It is an overly simplistic analogy but, at first sight at least, the integration of Lana Labs into Appian as the alter ego of its workflow tools could be the way that business process orchestration right out to the edge becomes a mainstream toolset for many businesses.