To Bot or Not? The Rise of AI Chatbots in Business

By Kieran Alger, Contributor

In 2017, chatbots are everywhere. They’re your personal shopper; they send notifications about your pizza order. Chatbots can update your bank account balance or serve as your hotel concierge—and this digital-transformation trend appears to be on the rise.

According to CITI research, Facebook Messenger chatbots are developing 70 percent faster than iOS apps were at this same stage in their life cycle. Up to 80 percent of survey respondents said they planned to use some sort of chatbot by 2020—and for good reason. In July of this year, Juniper Research forecasted that businesses using chatbots could save $8 billion per year by 2022 in banking and healthcare alone.

But chatbots have not only captured the attention of brands and their eager consumers—improvements in natural language processing and artificial intelligence (AI) have also created a fertile ground for chatbot technology in all business sectors, including B2B.

With the rise in chatbot adoption, the question for many business leaders is: How do I know if—and when—I should capitalize on this digital trend?

Chatbots Across the Organization

Early chatbot adopters zeroed in on customer service as a prime money-saving feature. In fact, Gartner predicts that by the end of 2017, only one-third of customer service interactions will require human interaction.

Yet despite this early focus, chatbot capabilities have broadened beyond customer support and expanded within—and across—organizations.

Today, chatbots make it possible to give every employee a personal assistant to help with tasks such as booking business travel, scheduling meetings, and managing to-do lists. Enterprise-grade bots can automate time-consuming tasks such as ordering supplies, paying vendors, and invoicing clients by using a messaging platform to manage these activities.

Departments also rely on chatbots as part of their larger digital transformation strategy. For HR teams regularly bombarded with employee questions, chatbots can seamlessly address queries with access to internal data. In one instance, Overstock replaced an employee hotline with a chatbot named Mila that manages employees calling in sick, giving their managers a heads up.

Bots can help internal colleagues and teams organize and make sense of vast databases. In the UK, Lloyds Bank created a chatbot to help staff navigate the organization’s entire knowledge base. “Chatbots actually for me have a bigger power, not in the customer-facing side but on the colleague-facing side,” Marc Lien, director of digital development and applied sciences at Lloyds, explained at London’s AI Summit.

To Bot or Not?

The question of whether and when to implement a bot will likely come down to its potential to streamline customer service, business processes, and data management. Below are just a few further qualifying questions top executives have begun to ask themselves.

Does my budget fit what I’m trying to achieve?

Chatbot pricing varies wildly, depending on the complexity of the system. A more simple text-based customer Q&A, for example, can be cheaper to implement than an artificial intelligence system that aids users with bookings. According to Yoav Rimon, a former Google employee and co-founder of Chat Leap, a marketing automation platform for instant messaging, there are three main pricing avenues.

In-house development, where internal teams design, build and launch chatbots, is likely the mostly expensive option due to employee, development, and maintenance costs, but it also enables deep customization using tools such as Microsoft’s natural language processing resource LUIS. Rimon suggests this is only advisable for advanced companies with the right infrastructure to support a chatbot team.

Self-service platforms, such as API.ai and Wit.ai, come with a lower pay-per-use model that services as an ad-hoc bot feature.

And in the middle are managed services solutions, like NuanceIPSoft, and Twyla, which provide support and technology and are able to integrate into existing enterprise data systems.

Are bots sophisticated enough for our needs?

Negative customer experience is possible in person, over the phone, or via bot, which is why it’s important to consider if chatbots have the sophistication to meet specific customer needs.

In natural interactions, people change their minds, use sarcasm or slang, and switch between languages. Most chatbots lack the ability to carry on truly personalized, human-like conversations that people may expect, which explains why Facebook Messenger bots only have a 30 percent success rate responding to requests without human intervention. While natural language understanding technology is improving, chatbots are currently best reserved for structured or scripted interactions, and businesses should avoid getting overambitious with the scope of chatbot functions.

Do I have the right security practices in place?

For chatbots to do their jobs well, they must collect and store data. Organizations must put fundamental security processes in place that govern where data is stored, who is using it, and what they are allowed to use it for. Employees must also be trained in chatbot protocols to guard against phishing scams. In particularly sensitive industries, chatbot communication should be encrypted.

Chatbots of the Future

There’s a lot to consider when it comes to implementing chatbots as a business tool. In addition to questions of quality and cost, leaders must fully understand organizational and customer needs before they land at chatbot technology as a solution.

But as we march into the final half of 2017, there is also plenty for decision makers to look forward to. Although chatbots have not yet peaked in their sophistication, they are proving to be valuable across organizations. Their potential to develop and streamline company processes—like the language-learning process that powers them—is limitless.

Much like the Internet 20 years ago, we’re on the brink of a tech disruption that may soon feel like a part of our daily lives.