With the possible exception of blockchain, can you think of a technology with a more turbulent ride on the hype cycle than Robotic Process Automation (RPA)?
Given our Corona-predicaments, our obligation to wade through the RPA hype has jumped through the roof. Remote work and emergency conditions have pressed the need for all types of robots and workflow automation.
But my lingering question is: what are the exact use cases - and what are the pitfalls? Coincidentally, I just experienced another rarity: an incisive PR pitch, this one from Nintex:
More businesses are looking to bring RPA into their business processes, but where do they start? Enterprises are in the dark about how to integrate bots into their existing processes, and those that go into the buying process with tech tunnel vision will meet roadblocks.
The RPA brochure forgot project realities
My RPA biases:
- I could care less how "robotic" the automation is. Companies have been automating workflows with varying degrees of success for decades. I care about the design - and proper integration with human escalation.
- Grilling practitioners at events convinced me that RPA can be effective - but it's also a heck of a lot harder than the RPA brochure. Yes, the technology is mature enough for some interesting use cases - but it's most certainly not a tech project. It's a business process project.
RPA use cases are popping out:
- The U.S. General Services Administration is using a bot to speed up collection of infection count data.
- The Wall Street Journal reported on hospitals using RPA to track COVID-19 tests.
- Kryon Systems shared several of their use cases in production, including the relocation of a global call center, health care data upload automation, and the automation of urgent customer outreach, e.g. banks dealing with an influx of delayed mortgage payment requests.
However, not all of these implementations are going swimmingly. The Horses for Sources team recently critiqued a problematic implementation by the U.S. Small Business Administration.
Nintex on RPA use cases - "businesses underestimate the implementation time"
Enter Chris Ellis of Nintex, who agreed to wade through my my RPA skepticicm. Nintex pitches itself as
"The market leader in end-to-end process management and workflow automation. Easily manage, automate, and optimize your processes with no code."
That's a pretty smooth pitch - perhaps too smooth for this grouch - but what about project realities? So I asked Ellis, who is Nintex's Manager, Tech Evangelism, APAC region, What are the biggest mistakes enterprises make when implementing RPA? Ellis responded:
- Most organizations considering RPA don't have a firm use case or pain they are looking to resolve with RPA.
- Much of the interest has been driven by RPA buzz, and the notion of 'we need to have it because everyone's telling us we should,' rather than looking to address a specific pain or problem.
Project scoping is a frequent problem. So is incorporating the human element. Ellis:
Businesses underestimate the time it takes to implement, and I don't just mean from a 'creating the bots' point of view, but also training the bot to handle exceptions when it comes to changes in underlying applications or sites, popups, errors and sanitizing data.
Customer example? Ellis says pushing data into a "single source of truth' is a popular use case:
I'm seeing a trend in RPA's ability to collate information across many disparate and disjointed data sources from local folders to cloud storage to email, and bringing the information into a single source of truth.
How about a remote work RPA scenario?
Invoice sanitization seems to be a hot topic for this particular example. We saw one Australian council replace a manual folder, email, scan and print solution with RPA and OCR, before passing invoices on for a 'four eyes' review process.
The need to replace the existing manual process was accelerated due to the remote work requirement, as printing invoices meant they stay in an in-tray for internal mail, something which is now obsolete as everyone works remotely.
Another example was provoked by
Excel hell remote work urgency:
I was approached last week by a large mining company to verify if RPA would be a fit for part of their onboarding process where there is a break in the automation chain to allocate a security pass to the employee. The door pass register is held in Excel, so it required a manual update for each new user. We're able to leverage the Nintex Gateway to automate this task, calling a Nintex RPA bot to replace the human-centric 'to-do' task in workflow.
When you run into "sprinkle this tech on everything" marketing over-saturation, I like to ask the reverse. So what is RPA not a fit for right now?
This goes back to my first point about underestimating the time to implement where a target system is dynamic and changes. Sometimes those changes can be so wholesale and granular it's almost a requirement to rewire the entire bot. The more moving parts in a system, application or data source, the harder it's going to be to continuously learn the changes for that bot.
Ellis thinks that issue can be solved by integrating a digital process automation workflow tool, which essentially merges a "human-centric review-and-approval overlay" with RPA.
RPA tech is mature enough now: you don't get to blame the tech if your project goes south. But even a valid RPA use case can prove problematic in execution. As the Horses for Sources team wrote in SBA: Please stop spanking bots:
It's too easy to say RPA failed. It's more complicated than that. Really RPA was a blunt instrument here – for its speed to solution and ability to swiftly process loan applications and it worked remarkably well. Too well. It swiftly overwhelmed SBA's E-Tran system – which was doomed to be overwhelmed anyway by these unprecedented volumes in a short period of time. But RPA exacerbated it and potentially gave access advantage to automation-savvy firms...
Automation always lives in an ecosystem with upstream and downstream impacts and these were not adequately addressed.
I don't know about you, but I find this type of sensible project talk encouraging. I worry about the phrase "Robotic Process Automation," as I think it somehow implies that there is some type of advanced machine intelligence going on with these projects. The vast majority of the intelligence in today's "intelligent process automation" is being provided by the humans that design proper RPA scenarios.
And here, I think we're all still on a learning curve. Example: I don't care for the type of chatbot on the Nintex home page, which forces human visitors who want to interact to choose between a set of brand-defined, limited responses. As a result, I get annoyed by another "dumb" bot, and the company doesn't get to find out a journalist was on their page, and what that journalist was looking for.
If an RPA firm is still working this out, I'd say we have a long way to go. But in the meantime, helping quarantine-challenged businesses with process automation is about as worthy a business project as I can think of.