The horizons of artificial intelligence keep stretching from robotic processes to ever greater customer and employee engagement through chatbots, virtual assistants and online and mobile capabilities.
To help clients in their digital HR strategies for workforce health, wealth and career, even firms that offer digital transformation services may need to create their own data strategies.
Cognitive automation (CA) is a set of technologies and tools that can take business capabilities to a new level by enhancing the functions and accuracy of business processes that rely on ever increasing data loads.
One way to do this is with a “technology garage” strategy, unpacking and applying the latest technology and design thinking with clients, experimenting with it in-house, like a venture-funded startup, and co-creating with vendor partners and startups along the way.
This reflects the recognition that the future of HR technology must be built through collaboration, teamwork and the launch of data platforms and microservices, as well as partnering with business leaders and identifying pain points.
These platforms enable data scientists to interrogate data for valuable insights that allow better workforce decisions though a deeper understanding of what the data reveals. One result that emerged at my firm was the Mercer Data Collector, the first online data collection platform for global, multi-industry HR survey participation.
How Cognitive Automation Can Be Deployed
Some key learnings emerge from the deployment of cognitive automation, focusing on three areas:
- RPA (robotic process automation), a cost-efficiency program to help automate various repetitive back-end administrative processes, freeing employees to attend to more complex tasks
- AI (through machine learning algorithms and data and analytics) to help gain insights and offer solutions to business problems
- NLP (natural language processing), conversational systems and virtual assistants to help engage with employees and clients
Applying AI and machine learning to client and business-process pain points begins with establishing a data platform—think of it as a data ingestion project to connect data across external (i.e., from social media) and internal systems (like CRM), along with algorithmic services to connect to and enrich the automation processes.
This resulted in several predictive algorithms for client retention. Financial and health-related algorithms were also developed through partnerships with external vendors.
As for NLP, conversational systems and virtual assistants like Amazon’s Alexa and chatbots can be extremely useful and efficient when engaging with customers or employees. For example, a service center chatbot can perform simple transactions or fulfill requests for information by walking a customer through a change of address or answering employee questions around open enrollment. Automating large volumes of simple and repetitive tasks can lead to big gains in operational efficiencies and free up employees for more important tasks.
Knowing the Problem You Want To Solve
Technology is a tool to be applied to real business challenges and opportunities to create value. Agile methodology and design thinking can co-create hypotheses and concepts to be proved. A healthy concept pipeline is important.
For example, design thinking innovation sessions with clients can help identify customer pain points. Innovation contests with vendor partners can crowdsource ideas and conceptualize them into potential solutions, using an agile/metered-funding approach. This is a method by which the partners iterate the solution based on a set of key performance indicators, metering the funding for a specific project rather than building out costly mega-projects without concrete KPIs. It’s an opportunity to transform financial processes.
From concepts, it makes sense to iterate pilot programs. Indeed, it’s a good idea to test the concept using multiple technologies to see how CA would work and what value, if any, it would yield.
Cognitive automation can enhance the functions and accuracy of business processes that rely on ever increasing data loads.
The Value of Chatbots
Mercer’s benefit service center chatbot for open enrollment was the result of one such pilot program.
During the last open enrollment season, the self-service chatbot was rolled out to over 150 clients, has proven more than 80 percent effective in resolving customer inquiries and resulted in positive operational savings. But the chatbot’s scope was kept narrow, allowing for quick reactions to customer needs, while providing valuable self-service options, especially outside business hours.
Targeted experiments like this can help build an ecosystem of technologies, capabilities, frameworks and partners that, based on agile methodologies, can allow organizations to build, buy or co-create with partners effectively to grow the business. Ideally, an additional result would be an inventory of reusable components on a “digital shelf.”
Integration Can Be Challenging
Still, there are some challenges that come with adopting CA—not the least of which is integrating it with existing systems. The technology and the expertise to run it is in high demand and can be challenging or difficult to acquire or find. And mapping old processes to new ones with enough detail to automate and addressing culture change management can be trying, so the organization’s subject-matter experts are critical.
To bridge that gap, classroom training is vital, using curriculum from an outside vendor and internal experts to ensure successful change management.
Educating colleagues about CA and process automation can help them make the transition smoothly. As a bonus, enhanced digital literacy leads to colleagues contributing new ideas to advance the business, creating an organic pipeline of ideas for new initiatives—a great example of the potential of the partnership between humans and technology.
As organizations push against the edges of innovation, they often come to realize that there are ethical boundaries. Algorithms, of course, are written by humans and are therefore subject to unconscious biases by their creators, which can skew the algorithms’ predictive effectiveness as they may apply, for example, to gender or ethnicity.
Establishing a data governance and quality process to capture and address those biases at the earliest stages of the algorithm-building process is crucial.
It may be complex, but it can be done. Indeed, the CA journey begins by exploring operational efficiencies and expands to more strategic programs that work to drive revenue or customer experience.
Doing it well calls for establishing a core set of frameworks and design principles, as well as educational tools to help the human element along the learning curve of change management. It may take time, but what begins in a technology garage can be rolled out for a great digital journey, powering organizations to successful heights.