Enterprises, Escape Your RPA Pigeonhole Today to Achieve End-to-End Automation

Enterprise customers want to use intelligent automation (IA) to improve how they run their businesses—usually by reducing manual work. HFS Research’s studies have produced evidence that shows that while robotic process automation (RPA) can help with some labor reduction, enterprises are learning that they can automate more processes with better accuracy and ROI by including cognitive elements to complement RPA, moving closer to HFS’ Triple-A Trifecta framework. We have begun referring to these emerging permutations of RPA plus another AI or analytics technology change agent as RPA++.

Enterprises may agree that RPA needs (elements of) AI such as natural language processing (NLP) or computer vision to enable end-to-end intelligent automation but bringing these change agents together effectively is still more art than science. There are no off-the-shelf tools that deliver broad-purpose intelligent automation.

HFS interviewed enterprises using RPA, service providers delivering RPA services, and RPA software companies to triangulate on the best ways to combine RPA and cognitive technologies to deliver advantages over standalone RPA. Ultimately, there is no singular winning strategy, but there are some playbooks that enterprises should examine to achieve RPA++. This POV discusses these emerging approaches.

RPA software vendors are making a concerted effort to complement their core offerings with cognitive capabilities

RPA software companies agree that RPA + cognitive is the future their companies are embracing— the method of the “+” varies. Some examples are:

• Some are adding AI functionalities to their core RPA products. WorkFusion was arguably the first to combine RPA and AI with its “smart process automation” capability. Other subsequent examples include Automation Anywhere with its ML-infused IQBot, Blue Prism’s newly announced AI Lab to develop proprietary RPA-ready AI elements, and AntWorks’ embedding of computer vision and machine learning in the front end of its IA stack to enable the use of unstructured data. This close integration means offerings will have more advanced capabilities from the outset, but, as AI is trained for specific tasks, it will also limit the scope of any hybrid solution and make scaling challenging.
• RPA software vendors are creating marketplaces enabling clients to connect to a wide ecosystem of partners and IP contributors. This brings AI functionalities into software vendors’ offerings without requiring expensive training for AI professionals in-house or developing AI tools from scratch. Examples include UiPath’s Go! marketplace and Innovation Fund, both designed to give UiPath clients access to best-in-class AI tools, Automation Anywhere’s pioneering marketplace, and Blue Prism’s equivalent.
• Other software vendors are taking a broad platform approach to combine RPA and AI. Thoughtonomy’s SaaS-based platform combines RPA plus an AI-enabled digital workforce manager (IADA) to manage and orchestrate automation. PEGA, through its acquisition of OpenSpan, added RPA capabilities to its AI-enabled customer engagement management to more broadly enable automation. The platform approach enables modular options and pre-integration, but the need remains to ensure there is a distinct focused purpose; otherwise, there is a risk of enterprises not knowing what to do with the platform package.

Service providers are taking a bird’s eye view of automation to tie RPA and cognitive together for clients

Service providers are doing a lot of the heavy lifting behind the scenes, creating rigorous methodologies to string together disparate technologies to solve particular client problems and developing their own IP that combines elements of RPA and AI. Examples include:

• Genpact’s Cora, a modular platform of digital technologies, similar to HFS’ Triple-A Trifecta, designed to help enterprises scale digital transformation through integration, orchestration, and governance of technology change agents.
• EY Lighthouse, a proprietary analytics-driven forecasting solution that combines elements of RPA and AI to minimize forecast errors (powered by Microsoft, delivered by EY).
• WNS’s TRAC methodology, “technology powered by robotics, analytics, and cloud,” keeps the company focused on embedding emerging cognitive technologies into RPA offerings to augment and supplement its clients’ automation deployments to help continually optimize business operations.

Enterprises are looking beyond RPA to true intelligent automation by thinking in terms of business outcomes

Enterprises are increasingly making efforts to combine RPA and cognitive elements into intelligent automation strategies, although it’s still early days, as shown in Exhibit 1. Often these efforts result in numerous POCs, most of which don’t go far, but it’s an encouraging indication of an important mental shift away from thinking of RPA as the holy grail of automation rather than a tactical tool in a bigger toolbox.

Exhibit 1: Enterprises’ current approaches to realizing intelligent automation

Question: How well are you able to develop integrated solutions leveraging multiple intelligent automation technologies to solve business problems?

^{Source: HFS Research in conjunction with KPMG, State of Intelligent Automation, 2019}

^{n = 590 business leaders, including 100 C-level executives}

Recent enterprise examples of evolving IA strategies encompassing RPA + AI include:

• After successfully scaling RPA in its customer service division and making it a standard component of its operating environment, Deutsche Telekom is starting to look to additional AI components to complement and extend RPA, as discussed in a recent webinar with HFS. Specifically, it is complementing its RPA base with packaged AI capabilities like cognitive assistants to enable broader automation.
• MetLife is reinvesting tax credits to create its Workforce of the Future Development Fund, which includes an intensive reskilling program for its employees in technologies including RPA and AI to actively prepare its employees for changing jobs and roles requiring IA. Having the right skills to enable IA is a critical path capability.
• Ericsson set specific financial goals for its automation overhaul. In its search for ways to hit these ambitious goals and derive more ROI from technology implementations, Ericsson found it necessary to involve more than just RPA, most notably AI and smart analytics.

There are still practical barriers to enterprise-scale of end-to-end automation, but they’ll soon start disappearing

The elephant in the room, of course, is that even AI cannot yet deliver full end-to-end automation for enterprises. Directionally, the future is RPA + AI, but practically, there is still no broad-functionality RPA++. This is in large part because current AI systems require intensive training and specialize in fairly narrow tasks—not unlike RPA. Moreover, the training data required to thoroughly train an AI model is either expensive or not in an accessible (structured) format, even within enterprises. 

However, this is changing as developments like memory networks promise to turn today’s laser-focused AI into general-purpose intelligence. As such, this gap is likely to start closing sooner than expected as memory networks lead to a proliferation of AI use cases. Likewise, the technology to make sense of unstructured and vernacular data is rapidly improving through the efforts of firms like re:infer.

The Bottom Line: Relying on RPA alone is not the answer. Enterprises must start looking at RPA as just one piece of the automation puzzle.

There is no doubt that enterprises have a duty to start planning how to position RPA within their automation agendas as just one cog in a larger, more complex machine, turned by more powerful and adaptable cognitive change agents, with AI at the forefront. The evolving options to access this broad range of capabilities include buying directly into hybrid solutions, adopting platforms that stitch AI and RPA together, and relying on a service provider to bring the right combination that meets the enterprise’s needs. The path, or combination of paths, enterprises take will depend on the clarity of the organization’s automation vision and its existing technical knowledge and resources to successfully implement RPA++. The time to start evaluating which of these approaches is right for your enterprise was yesterday, so there’s no time to waste. Success favors the bold and the fast!