I recently came across an article from ZD Net that contained an interview with Stephen Pratt, CEO of AI solution provider Noodle.ai.
In the interview (“AI and the enterprise: The view from Noodle.ai”) published on November 15, 2017, Pratt argues that many companies are starting their AI initiatives in the wrong place—chatbots. Instead, he contends, they should be focusing on enterprise-wide issues such as the business planning process and core business operations.
These are admirable goals. However, the types of AI applications Pratt is suggesting, while valuable to the enterprise and solving real and difficult business problems, are somewhat inaccessible to rank and file marketers, sales leaders, customer service organizations, front line department managers and other knowledge workers whose jobs could also be enhanced and made more productive with AI.
Chatbots, on the other hand, are much more accessible and tangible to the typical business person, and many of the people in customer facing roles immediately understand their potential. Bots can improve engagement and reduce customer service costs (if correctly architected of course). They can serve as the front line interface, solving the more straightforward problems, leaving customer service representatives free to take on and resolve the more complex issues.
A number of challenges arise in the case of the more sophisticated AI applications. First of all, they are more difficult, time-consuming, and expensive to develop. The work required to create an intelligent supply chain or complex resource management application can become an enterprise-wide initiative that can easily get bogged down. It takes time to muster the resources, whether internal or external, needed for a large implementation. If the project is too ambitious and does not succeed, the organization might end up wasting resources, or be deterred from using AI in the future.
Pratt contends that organizations are better off putting their efforts toward higher value AI applications such as the supply chain optimization tools that his company provides. While there may be an excellent ROI for that class of tool, chatbot applications can also produce a solid ROI. No matter which approach a company takes, the point is to carefully consider the use case, be realistic with expectations, manage scope and deploy rationally, measure results and improve incrementally. Whether supply chain applications require heavy duty programming or are configured through point and click interfaces by business people, the way to look at artificial intelligence is not “as an AI application” but rather as “an application that uses AI,” Considering chatbots as an application that uses AI, the application is more about the data than the algorithm.
Chatbots are a channel to data and information. They leverage machine learning and AI, and require the correct data – in this case curated content – in order to function. Chatbot content is targeted to a domain and specific process and is correctly tagged for retrieval by the bot to meet the needs of the user. By virtue of this focus on process, content and user needs, many organizations are gaining new insights into how they serve their customers with information. This is causing greater resources and attention to be devoted to the nuances of the customer relationship which is already paying off for many organizations.
Chatbots can be a good place to start for companies that are not yet ready to dive wholeheartedly into AI on a grand scale or whose priorities are related more to customer engagement and experience than supply chain optimization. Chatbots can work effectively to further such objectives as personalization and provide data for predictive analytics that support other corporate strategies.
I agree with Pratt that many organizations are embarking on ill-conceived AI initiatives and that chatbots can be like a piece of crumpled aluminum foil to a squirrel – a shiny object to play with that does not provide value. Many business and technology leaders can be easily distracted by shiny bits of foil, and vendors are masters at hitting the hot buttons of leadership by articulating how their tool (their piece of foil) solves pressing business problems.
As I mentioned, part of the attraction of chatbots is that they are understandable to most non- tech people. AI concepts and applications can be difficult to grasp by even the tech savvy. Chat is part of our everyday experience and chatting with a bot can provide immediate value by allowing users to complete clearly defined tasks with minimal effort. The value of a chatbot initiative is the understanding the organization gains about what the user is trying to accomplish and what is required to provide a low friction path to achieving the goal. This is the goal of a seamless customer experience and it is more difficult than most realize.
Bots can reduce support costs, improve the customer experience, engage throughout the sales process, provide entertainment, and strengthen the brand by providing targeted content and real-time data. The goal is to offload tasks that are either handled by a human or that require the user to sift through content and navigational menus to solve their problem. Bots can (and should) be designed to serve very clear objectives and produce unambiguous results that provide tremendous value to organizations.
One of the more interesting side effects of bots designed to improve the customer experience is the increasing attention that organizations are paying to their unstructured data and content. Data hygiene is key to AI, and content hygiene is key to chatbots. The training content that-AI driven chatbots require is in the form of knowledge assets and curated content. Chatbot development is also requiring organizations to focus more attention on user needs and to the precise information they need at each stage or their journey.
AI and predictive analytics are extremely valuable in smoothing out the supply chain by predicting demand, as Pratt explains, and these tools are examples of predicting market needs and consumption patterns at a systemic level. AI drives web personalization of offers and presentation of content, which is a similar predictive pattern mechanism at a micro level – that of the individual.
Bots are an extension of this personalization– they understand what users want and need, and predict what will satisfy them, whether that is an answer to a question or a particular product in the case of conversational commerce. At this micro, individual level, AI-driven chatbots are prediction engines that run on the same principles of responding to signals as supply chain optimization engines. So bots are a logical extension of AI approaches for serving customers more effectively, which is the reason we optimize supply chains in the first place. Therefore they can serve an important role in establishing AI at the grass roots level, and nurturing the groundwork, including developing good content curation and data hygiene practices, that will prepare companies for more advanced AI initiatives,
To summarize, both the sophisticated AI applications and the more targeted ones such as chatbots have their merits. There is a lot of hype about chatbots and in order to be useful they need to be developed and used judiciously. However, they should not be dismissed as an entrée into AI, a tool to catalyze better data management, and a way of gaining understanding of customer needs.