Part one of our four-part IRPA AI Ecosystem Webinar Series, “From Hype to Reality: Man & Machine.” You won’t want to miss a single event as we travel the RPA road ahead together and explore this new landscape and expand our horizons. We have an exciting lineup of quarterly presentations and discussions from key players from the buyer, provider, advisor, and analyst community. They will share their insights and observations on the impact RPA, AI, cognitive computing, machine learning, and intelligent automation are having on today’s business.
The Road Ahead – The Technology Landscape
In part one of our series, we examined the different kinds of automation technologies and discussed what they are, how they are used and how they are maturing and evolving. Adam Devine, Vice President & Head of Marketing, WorkFusion and Paul Pinto, Managing Partner, Sylvan Digital led this highly interactive discussion. See the on-demand webinar below.
Questions from Audience:
Adam…Great Presentation…how did the BPO provider respond to your SSI example?
ADAM: BPO providers are generally responding to the customers’ use of WorkFusion in one of two ways: by becoming a WorkFusion customer and proactively automating the repetitive manual work that their workers are doing and sharing the cost savings with their customers, or by losing contracts. We see more BPO providers seeking to partner with us to serve customers than the latter.
Are there any deployments that workfusion has done in India?…Does their pricing make a good business case in the Indian context – given the exchange rate?
ADAM: WorkFusion is delivering two deployments right now for two global banks and several global BPO businesses are being trained on our software as well, which we expect will lead to many other India-based deployments for both end users and service providers. Our pricing reflects fair market value for the capabilities provided.
What is the typical trigger point for the automated processes – e.g. does WFusion address automating use cases like Infrastructure monitoring and automated resolution of system events/incidents?
ADAM: WorkFusion is focused right now on business process automation rather than IT process automation. That said, smart automation is typically triggered by an event (e.g., the mention of a keyword in a news article or the receipt of an email message to an inbox), a time (e.g., 8am = begin a process), or by a manual user prompt.
Is there any risk this new automation paradigm creates a bubble in the following years, similar to the one we lived in the last 90’s with internet business?
ADAM: The difference between the internet bubble in the 90s (which I was very much a part of) and smart automation is two-fold: smart process automation solves a real and valuable problem for customers, whereas Pets.com, theglobe.com, webvan.com and other similar web 1.0 flops were solutions to problems that did not exist or worse, were poorly constructed solutions to real problems and did not deliver. The difference is in the ROI. WorkFusion and other SPA vendors are delivering real and rapid ROI to very serious publicly traded businesses helmed by very pragmatic and discerning leaders. If the word “bubble” is at all relevant in the context of SPA, it’s that the outsourcing bubble for providers that let themselves fall behind in their use of SPA will burst.
PAUL: Because Smart Process Automation is truly a disruptive technology, it will most likely create several different types and shapes of bubbles. As a result of implementing Smart Process Automation, we are already witnessing the mass repatriation of services from offshore providers, as well as the creation of new job roles (e.g. Automation Architect, Automation Expert, Automation Curator).
What tools are currently missing or underdeveloped in the market today?
ADAM: There is a big difference between active learning and self-learning in machine learning, the branch of AI that WorkFusion leverages. Cognitive automation is powered by active learning, which is when a human performs work on a software GUI (graphical user interface), and algorithms track human actions through this GUI and essentially maps the unstructured content. Self-learning is when automation receives a new document or PDF or email message and, without any training, uses human-like judgment and subject matter expertise to perform the work entirely autonomously and uses this self-acquired knowledge to perform incrementally more challenging work. This is form of AI is still a ways off, but we’re working on it. Human-in-the-loop active learning is very much state of the art right now, and it enterprise proven.
PAUL: This technology has been around for quite a while (10+ years) in various forms, and only recently (within the last 3-years) has it started to be applied in ways that have a meaningful impact on business. As such, there a variety of tools that support Scripting, Robotic Process Automation, Autonomics, Cognitive Computing , and Artificial Intelligence, with certain tools focused on automating business processes, while others are better tuned to support the automation of I.T. processes.
What are some of the apprehensions that you have seen with clients and do you think more visibility and education around smart automation would assist in relieving those apprehensions?
ADAM: Every single customer in every industry is at some point concerned about security and data quality. Us assuring them with facts and case studies certainly helps, but the quickest way to educate is first-hand proof. Several of the biggest banks in the world overcame their security and quality concerns by running their existing processes and WorkFusion automated processes in parallel and found that a) security improved due to less exposure of sensitive information to human workers at the desktop level and b) quality improved thanks to automation accuracy + human exceptions processing.
PAUL: As with the introduction of all new technologies, User’s always seem to feel that their environments are too complex to be automated by a tool, and while these same Users are receptive to further education, they still cling to the comfort afforded by their convoluted infrastructure. As such, It has been our experience, that the most productive way to further the perpetuation of Smart Process Automation is to actually automate a complex process for the Users. By delivering this “Proof of Value”, the Users can be relieved of their historic prejudices and enabled to become advocates of Smart Process Automation.
RPA to Autonomics can be handled as an in-house solution, much quicker and efficiently than vendor based solution, yet it would appear that internal culture issues prevents this much of the time. How would a Consultant help the Enterprise leverage its own capabilities?
ADAM: RPA is easy. If your organization has the IT time and budget to script its own bots, godspeed, but if there exists an enterprise anywhere in the world that has developed a viable inhouse cognitive automation solution, please email me directly at [email protected] so that I can get the creator’s info. I’ll send him a bottle of Dom with my congratulations for achieving what took us nearly 5 years, some of the smartest data scientists and engineers in the world, many IP patents, and a lot of venture capital investment.
PAUL: While there are many good reasons for engaging a knowledgeable and experienced third-party Consultant to assist in the introduction and instantiation of any new technology, with specific regard to reaping the benefits of Smart Process Automation, it all about human change management. It is certainly true that any mature I.T. organization will inherently possess the capabilities to learn and implement Smart Process Automation tools, but this isn’t just about the implementation of technology. The success or failure of Smart Process Automation has everything to do with End-User adoption by Business Users, and very little to do with I.T.
I have seen end to end RPA, i.e. what you are calling Autonomics or other call Augmentation. Now I have moved on to what you are calling Cognitive Automation. Is this a logical evolutionary process, or can you just jump to Cognitive and then full AI. I see it logical evolving from one to the next. Best I have heard!!!!
ADAM: Thanks! Robotics are still useful for very basic desktop UI automation. Humans don’t need to train automation to enter a password into Concur or to open PeopleSoft. Think of automation in terms of the automobile. You could still use a Model-T to visit the grocery store, but it would be ill advised to take it on the highway. Another example: you could use a sushi knife (cognitive) to cut both butter and sushi, but you cannot use a butter knife (robotics) to cut sushi.
PAUL: It is fair for us to view the five stages of Smart Process Automation (i.e. Scripting, Robotics, Autonomics, Cognitive, and Artificial Intelligence) as being evolutionary, because the thought of one stage building on the foundational thoughts of the previous stage is a comfortable paradigm. In reality, evolution is occurring in-parallel, within each stage, independent of the other stages. The field of Artificial Intelligence, with it use of Neural Networks, is very different from the pattern recognition capabilities within Cognitive Computing, and even more different than RPA’s machine learning.