The word “Chatbot” is officially banned: they treat conversations like they’re a game of tennis: talk, reply, talk, reply. There is little to no context and zero intelligence, just pre-programmed responses only set up to deal with a pre-set finite number of frequently asked questions. It’s a legacy customer experience that most of us go out of our way to avoid. To be blunt, it’s easier to be redirected to an FAQ page, or even some online Q&A forum than try and engage in a dumb one-dimensional conversation. I’ve had more intelligent conversations down my local pub after a 3.00am “lock-in”… So let’s shift the entire conversation towards chatbots with some form of intelligence…cognitive assistants.
HFS Research sees cognitive assistants as the combination of conversational interaction and process execution capabilities; they combine characteristics of smart analytics and artificial intelligence. These services can include front-office facing elements (e.g., conversations with end customers) and internal employee use cases (e.g., help desk, HR onboarding, assisting contact center agents).These cognitive assistants can self-learn, self-remediate, and execute business processes. They can also often understand structured and unstructured data and then use natural language processing to learn, comprehend, and recommend next steps. Advanced cognitive assistants can also enable predictive decision making using real-time analytics. This distinction is significant as many people use the terms “cognitive agents” and “chatbots” synonymously. While cognitive agents are a less mature capability, interest and adoption are growing rapidly—and their impacts are far greater than traditional automated tools.
So who’s delivering these services most effectively today? Well, who better to consult that HFS customer experience connoisseur Melissa O’Brien, who’s just launched the industry’s first deep-dive report on the services market for these cognitive assistants:
We based this research on interviews with 300 enterprise clients of IT and business services from the Global 2000 in which we asked specific questions about innovation and execution performance of service providers assessed. We augmented the research with information collected in Q1 and Q2 2018 through provider RFIs, structured briefings, client reference interviews, and from publicly available information sources.
Melissa: Which service providers are the early front-runners?
The most notable players in this study demonstrated the most sophisticated and well-developed cognitive assistants. These providers’ services had solid examples of use cases with capabilities above and beyond the traditional rules-based chatbots and IVR style communications. These intelligent bots have the capability to learn, handle unstructured data, solve business problems and use natural language processing to analyze sentiment and understand context, and are deployed across the enterprise in a variety of ways.
IBM, which topped the list of providers demonstrated the greatest volume and depth of cognitive assistant use cases across industry verticals and enterprise processes, and has the power of Watson as a brand and a tool to leverage for these services. Cognizant, the runner up, also has a wide variety of deployments and some unique examples outside of traditional channels, like its cognitive assistant in a kiosk of a quick serve restaurant. Accenture had some very strong virtualized deployments and the most sophisticated consulting and design capabilities for cognitive assistants.
Which other providers are also showing potential?
One service provider with a unique focus is CSS Corp, which has doubled down on its “Yodaa” cognitive assistant capability. While it risks getting pigeonholed by going to market with the one persona, its actual use cases are broad and the sales and marketing outcomes achieved in particular have been significant. TCS demonstrated broad applicability for its cognitive assistants services, with significant depth and scale of deployments and compelling and differentiated use cases in particular for travel and the NYC Marathon (check out “Pacey Miles”. Many of the service providers, including TCS, Accenture, Infosys, Tech Mahindra, Concentrix, Sutherland, and Convergys, have developed cognitive assistants for their internal processes—using them within their own HR and IT processes or to assist their customer service agents — which is demonstrative of their faith in the solutions they’ve developed. Wipro, DXC and HCL have each demonstrated strength in deploying cognitive assistants within clients’ IT service desk environments— automating ticketing and issue resolution in a more intelligent way.
Is the focus of cognitive assistants is more about augmentation of employee work rather than replacing them?
Automation tools can often replace a human interaction—we see this a lot in self-service, especially in the case of straightforward, focused inquiries. Tools can typically free the employee to do something less transactional, more valuable to the customer, and more “human.” However, with cognitive assistants, the capabilities are more powerful and therefore more nuanced. Generally, the use cases we’ve seen are about making employees, whether contact center representatives, IT service desk staff, or human resources officers more efficient and effective; often that means that the bot is working side-by-side with the employee as an assistant, synthesizing and presenting data, aimed at making their lives easier and processes more intelligent and agile.
Which processes are more commonly being used for cognitive assistant deployment?
Front-office deployments are common, but their AI implementations are not as mature as examples often found in HR, finance and accounting, and help desks. The majority of case studies we saw in this research involved the front office, particularly in sales and customer service. These are often the starting points or the low hanging fruits where enterprises will decide to test the use of cognitive assistants. But the capabilities for cognitive assistants go well beyond the front office, assisting in various elements of the enterprise such as HR, finance and accounting, and the help desk. While the front-office examples are ubiquitous, more mature use cases are often found in other areas where cognitive assistants can execute on processes such as ordering equipment for an employee during onboarding or creating and resolving a help desk ticket autonomously.
Will this all boil down to smart partnering, or a best-of-breed approach?
Partnerships are essential building blocks for cognitive assistants. Many of the service providers in this study cited a “unique” approach with “best-in-breed” technology providers. The reality is that the technology is advancing so rapidly that there’s really no such thing as best-in-breed, and having a partner ecosystem is hardly unique. Those leading in this market will develop strong relationships with well-known players (e.g., IBM Watson, IPsoft’s Amelia, Nuance for NLP), which is essential to have a flexible client-friendly environment—but will keep a keen eye on up-and-comers. Integration with other systems (e.g., ServiceNow for ticketing, HCM platforms for recruitment and onboarding, or CRM systems for customer data) is also important. Almost all of the service providers we spoke to have a technology-agnostic platform (perhaps with the exception of IBM, which partners but leverages the Watson platform heavily), which enables them to leverage their clients’ existing investments and be flexible to clients’ needs and modular with building the tools.
If this market is so focused on the front office, Melissa, why aren’t the contact center providers leading the pack?
Pure-play contact center BPO companies are less mature but have tremendous potential to move up the value chain. The contact center BPO companies (Convergys, Sitel, and Teleperformance) we profiled had less mature capabilities and fewer actual client case studies; two reasons are that contact center BPO companies are finding that it is difficult to fit cognitive assistants into their bread-and-butter business and that automating customer interactions brings with it revenue cannibalization. However, for front office use cases there is a tremendous opportunity for these players to take the lead given their wealth of customer data and customer experience expertise. By embracing cognitive assistants, these service providers have the opportunity to carve out a differentiated capability for a blended bot and human model, providing seamless transitions to human agents and harnessing the power of their core capability—while potentially breaking out of the legacy FTE models that have dampened innovation and profitability for years. Two ripe areas for further developing cognitive assistants for contact center companies are in use cases that employ bots internally for recruiting and hiring and those that augment agents. Companies that use these tools internally to their best advantage will create differentiation in their service delivery.
The Bottom-Line: The disruptive potential of cognitive assistants is only just unravelling for service providers and enterprises
Today, the cognitive discussion is about augmenting peoples’ performances. Being able to engage in a semi-intelligent manner with a bot has huge ramifications for everyone in the business environment: for consumers who want to engage digitally, for employees who can spend less time having routine interactions, for supply chain staff who can move products and services around at digital speed, and so on. The common nuance here is that all these entities are being assisted – they aren’t being replaced.
Tomorrow’s discussion is about scaling businesses with less reliance on people addition, and these cognitive assistants will be a significant factor. However, the evolution of technology isn’t finite, and while these social tech tools today add value and efficiency and add to the customer engagement experience, the more these tools can be refined to self-learn and self-remediate, the more judgement-based human tasks they will be able to support. Ultimately, this means both service providers and enterprises will not need to keep adding so many new staff in the future to deal with larger business volumes. Cognitive technologies, when smartly managed, will allow business to scale without the linear addition of staff to deliver scale. This creates a huge opportunity for service providers to compete, as those which can provide volume services and lower rates, based on their acumen to deploy cognitive assistants to support their clients. Of course, enterprises will have the choice of deploying cognitive assistance themselves inhouse, or whether to engage with external service partners which can deliver high volume/ low-cost cognitive services for them. Much depends on the skills needed to develop ongoing algorithms based on an intimate understanding of the business and its institutional processes. This has much more far-reaching potential than merely deploying an RPA bot to crank some basic process loops… This means having unattended and attended interactions with people, processes and data sources both inside and outside of the enterprise, such as macroeconomic data, compliance issues, competitive intel, geopolitcal issues, supply chain issues etc.
Net-net, the more intelligent we become at deploying these cognitive assistant tools, the more we will focus on technology-driven solutions that reduce the numbers of butts-on-seats needed to support the business. Those service providers which fail to deploy cognitive assistants to scale their services competitively will fall away, and for those which have not yet developed this emerging capability, it may already be too late. I doubt we’ll even call these solutions “cognitive assistants” in a couple more years… they’ll simply become part of the fabric of operations, and how we engage and interact digitally to get things done. Rather like offshoring became so normalized we stopped defining and noticing it, the same will happen with many of these emerging technology solutions as they become part of how we conduct business.
(Cross-posted @ Horses for Sources)