Check out the on-demand sessions from the Low-Code/No-Code Summit to learn how to successfully innovate and achieve efficiency by upskilling and scaling citizen developers. Watch now.
Microsoft has announced significant advances to its Power Automate platform to help scale robotic process automation (RPA) infrastructure. Key advances include new capabilities for understanding business processes, collaborative bot development, and scaling RPA software bots with virtual desktops. Microsoft is a relative latecomer to the RPA playing field, but is growing this capability quickly — thanks to its existing strengths in office productivity apps, Windows integration, and Azure cloud infrastructure.
The field of RPA started as a way to program sophisticated macros for automating repeating tasks like copying and pasting data between two business apps. Gartner has suggested the future of this field, known as hyperautomation, includes finding ways to identify automation opportunities and program automations, then scale the deployment of the automation more efficiently. Microsoft’s newest updates tick the boxes of significant progress on all three of these aspects.
Microsoft’s Power Automate general manager Stephen Siciliano told VentureBeat, “Understanding up-front which processes have the most wasted time and the highest potential for automation will be extremely valuable for them.” His team aims to provide a single end-to-end product that improves the complete hyperautomation lifecycle and is deeply integrated into Microsoft 365 and built into Windows OS.
New process mining improvements
Many enterprises have hundreds or even thousands of processes that could be automated. The initial step in figuring out how to identify automation opportunities at scale lies in automating the process of seeing what procedures enterprises are commonly repeating. One approach called task mining watches over a shoulder to see how someone clicks and types their way through a process using a sophisticated macro recorder. Earlier this year, Microsoft provided the ability to bootstrap an automated flow from a task mining diagram, making it easier to go from seeing the process to optimizing it.
Intelligent Security Summit
Learn the critical role of AI & ML in cybersecurity and industry specific case studies on December 8. Register for your free pass today.
The second approach, known as process mining, analyzes enterprise application event logs to reverse engineer a process diagram. This is important, especially when a process spans many users and enterprise applications. Microsoft’s new process mining capability takes advantage of existing Power Query connectors for Power BI and Azure Synapse. So although the specific process mining aspect is new, Power Query supports data ingestion from the hundreds of enterprise applications today. In addition, Microsoft acquired Clear Software last week, which will improve connectivity with enterprise applications like SAP and Oracle.
Microsoft’s process mining and task mining capabilities approach visualization and the understanding of a process from different angles, which are slowly becoming more connected. Process Mining uses event data from systems of record to understand and analyze the process in a company. Task Mining fills in the gaps in the process identification which event data cannot see, what the human does in the middle, using recorders and RPA techniques.
“In the future, we anticipate that the lines between those two will blur more, and the focus will be on the full end-to-end process, no matter how many tasks or systems it touches,” Siciliano explained.
Collaborative bot development
A second significant advancement improves collaborative bot development. The process and task mining capabilities help generate a template for how a bot is supposed to click and type its way through a particular workflow. But humans must then look over these and identify how to design a more efficient process or respond to common problems.
Microsoft has added collaborative development workflows that allow different experts, including bot developers, business process analysts, subject-matter experts, front-line users, and compliance teams, to collaboratively make comments, recommend changes, and revert or accept the recommended changes. This takes advantage of the same commenting infrastructure used in Word.
One concern is that RPA bots could copy data to applications lacking appropriate protections or to physical locations in violations of regulations like GDPR on where data can be stored. New data loss prevention capabilities widen Microsoft’s existing capabilities for labeling and tracking sensitive data into RPA automation.
Historically there have been two sets of machines that enterprises have to manage for RPA: the servers that orchestrate the work that needs to happen and the machines the bots run on. Power Automate automatically manages the machines for RPA orchestration in the cloud. However, enterprises now must manage the machines that contain the running bots. Microsoft is previewing the Azure Desktop Starter Kit, which can automate this second aspect as well.
Siciliano said, “This will make it possible for IT to focus less on the raw tasks of setting up and scaling machines, and more on enabling more users in their organization to build out bots.” This could also simplify governance since the IT team can set the exact policies on its use and who uses it.
Better integration between RPA and Azure desktops promises to simplify the processes of setting up and scaling the appropriate machine configurations for RPA deployments.
“We’ve heard from customers today it can take days to weeks to bootstrap and scale infrastructure,” Siciliano added. “This means not only are citizen developers blocked (and thus not being productive) during that time, but also, there’s a lag time of machines sitting idle. All of that goes away once you can scale automatically.”
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.