There’s a beautiful plank of figured cherry tucked away in my woodshop that I’ve been saving for a special project, so I sat down at the workbench the other day with some tools and a set of plans. But rather than starting by milling the cherry to size, I spent a few hours building a jig – a hardwood and steel contraption that will hold each workpiece firmly and precisely while it glides past a router bit.
The cherry cabinet will be held together by box joints, identical little fingers that fit tightly together, combining a crisp appearance with excellent mechanical strength. There are many fingers on each piece of the case, so the joints will only fit together properly if they’re cut accurately, consistently, and without error. Naturally, this got me thinking about software bots.
Forrester Research recently released a report on the emergent field of robotic process automation (RPA) – software bots you train to recognize recurrent activities in your everyday apps and websites so they can perform those activities automatically. According to Forrester, by 2021 there will be more than four million robots doing administrative and sales-related tasks. Leading vendors in this area include Automation Anywhere, Blue Prism, UiPath, WorkFusion, Pegasystems, NICE, Kryon, EdgeVerve, and Redwood Software.
RPA is increasingly popular in the enterprise because – just like a woodworking jig – it lets knowledge workers perform repetitive tasks with increased speed, accuracy, and autonomy.
- Speed – Many routine corporate tasks, such as filing an expense report or processing an invoice, can be done more quickly when a bot recognizes and automatically copies data from relevant fields in a spreadsheet;
- Accuracy – Once you’ve taught it how, a bot will fill in a zip code, part description, or contract boilerplate with absolute precision every time;
- Autonomy – Bots can be configured by end users without programming skills, and because they interact directly with each app’s interface, a software developer isn’t needed to provide “under the hood” integration.
There’s one other way in which robotic process automation resembles a woodworking jig – it requires an investment in time and materials. Woodworking books and magazines are full of plans for finger-joint jigs, with many emphasizing how quick and easy they are to build. Instead, I started with a more complicated design, constructed from a piece of quarter-sawn white oak with heavy steel brackets. In return for a little extra time and trouble, I now have a jig that cuts airtight joints, and can be adjusted to within a few thousands of an inch, with absolute repeatability every time.
It’s the same way with RPA – product capabilities and prices range widely, from small companies leveraging an open source model, to large organizations building AI-enriched systems priced at five figures per bot. And as with any emergent technology, management and governance are growing concerns. Among the questions business leaders are asking:
- How can we capture knowledge gaps in our processes? Once a process has been programmed into a bot, what do we do when the person who really understands the process is no longer involved?
- How can we ensure compliance with internal policies and government regulations? Existing systems include rigorous checks and controls, many of which can be bypassed by bots that mimic a human’s interactions with apps.
- How can we centralize command and control? Especially as bots incorporate broader machine learning capabilities, how will we centralize control of work queues, execution rules, schedules and support, so we can spin up and down bots as required?
I’ll explore these questions, and some possible answers, in a future blog post.
Links:
- Forrester Research report (download from Automation Anywhere website)