RPA: Becoming the Linchpin of Enterprise AI
With the growth in robotic process automation, a new era for the contact center is upon us. This is big.
The number of enterprise customers for the top three robotic process automation (RPA) manufacturers, UiPath, Blue Prism, and Automation Anywhere, has increased from approximately 200 in 2016 to about 4,000 in mid-2018. Forrester Research estimates these companies have helped enterprises around the globe deploy four million "digital employees" -- bots, essentially. During this explosive growth, the industry has shifted from primarily supporting unattended automation (automation tools that don't require human interaction) to attended automation (tools that rely on human interaction). A new era of contact center RPA is upon us.
From a contact center telephony perspective, this means availability of a new method for automating customer interactions that has nothing to do with telephony. In other words, digital employees will soon make customer support interactions faster, more consistent, and less dependent on human agents. If your business is dependent on selling or supporting telephony-based customer interaction tools, then the next couple years will be tumultuous as contact center operators begin reducing their needs for human-operated, real-time customer assistance tools.
Forrester 's Wave market analysis, available here, provides a great overview of the RPA industry. I cover the high points of this analysis below, but first, a few words about architecture.
In the telephony business we have a long history of delivering IVR, CTI, and now customer journey analytics with the aim of reducing cost and improving the customer experience. Omnichannel support is now becoming more prevalent, and we're finding that automating text, chat, and mobile interactions is creating new challenges. Add natural language processing (NLP) to the mix, and the permutations related to the integration of back-end systems is becoming overwhelming.
The solution is to create a layer of abstraction between the real-time interactions systems and the back-end information systems. This is the architecture that the RPA business is exploiting.
A simple example of this architecture is a bot that retrieves customer information for presentation in multiple interfaces -- voice, text, mobile, etc. RPA allows every channel to use the same function, whereas a different set of integration tools for each media interface is necessary with traditional telephony solutions. As a contact center operations manager you have a choice to 1) follow the traditional model and witness a proliferation of custom-written information interfaces in the enterprise or 2) use an RPA tool to develop, enable, manage, measure performance, orchestrate, and maintain these automated information systems functions. Which would you choose?
Forrester calls out the following categories in its analysis:
- Performance reporting
- Machine learning
- Ease of deployment
- Attended/unattended automation
- Access to peripheral automation solutions
- Process orchestration
- Regulatory compliance
- Revenue/customer base
Interestingly, NLP is not part of the comparative analysis. I believe this is because every RPA platform in the analysis offers it. Further, most have been using NLP for years.
In its report, Forrester posits that you should manage your digital employees the same way you manage your human employees. It seems, at least for the leading vendors, that performance reporting is available on every bot. This reporting is much more granular than Average Speed of Answer (ASA), Average Handle Time (AHT), Abandonment Rate (ABN), and Service Level that we traditionally use in contact centers. This is where we can use AI, machine learning, and analytics to improve the customer experience.
Ease of deployment is probably the most profound change that RPA brings to the market. For several years now, RPA vendors have focused on automation of computer processes with development interfaces oriented toward non-developers. These interfaces mimic many IVR GUI development interfaces. They have done this by using the same "player-piano" techniques used in early CTI implementations.
For instance, an RPA build can provide the bot with access credentials for a CRM and mainframe solution, and then programmatically identify the content on the CRM GUI that needs to be copied. It would then use the same GUI technique to paste the content into a mainframe interface -- no coding skills required. In some cases, the RPA solution would let bot developers just point and click to identify the relevant, on-screen data that needs to be manipulated. At this point, the best bot developers are the contact center or back-office employees themselves.
The point about the bots' use of access credentials for a GUI interface when APIs aren't available will have a particular impact on industries that rely on "closed" systems and software. Users of the predominantly closed software that is available in the healthcare business will benefit from these tools in ways that the software companies have thwarted for years. Other opportunities are baked into custom-written software that often does not include APIs.
Process orchestration is another RPA platform component that contact center managers should explore. Many contact centers are challenged by business processes that require interaction between the contact center and back-office operation for reasons like research and adjudication. If you have work-queues in your operation, then RPA is in your future.
Security and regulatory compliance are baked into most RPA platforms. Masking of confidential or protected data and encryption are table stakes in the RPA business. Most RPA platforms allow the use of NLP in analyzing real-time or stored copies of interactions for regulatory compliance. In addition to regulatory reporting, these capabilities can be rendered for use as regulatory agent-assistance tools.
For those in the contact center operation business, a new wave of process automation is upon us. Its key attributes include AI, machine learning, NLP, easy implementation, advanced performance reporting, multitenancy, orchestration, security, and regulatory compliance. The reduced need for craft-oriented software development via improvements in GUI "development" environments make RPA more accessible than any other contact center automation technology. This is big.