Practically every business with a customer service department is thinking about bringing AI into the fold. According to Accenture’s Banking Technology Vision 2017 report, four in five bankers believe AI is poised to revolutionize the way financial institutions gather information and interact with customers and three quarters of those surveyed bankers see this shift happening within the next three years.

Of course, artificial intelligence is a large technology space comprised of different technologies that have been developed for a variety of purposes. Yet some AI technologies such as chatbots and virtual agents are frequently grouped together when in fact they couldn’t be more dissimilar. Each thrive in different business environments, however they address very different challenges.

Fundamentally, chatbots are not as functionally versatile as their virtual agent counterparts—primarily due to their cognitive limitations. Chatbots are generally built to respond to customer requests according to a pre-determined script. They are especially effective when the requests being asked by a customer are simple and do not require additional context.

This distinction, the use of context, is critical. Without context, a conversation is singular, or uni-directional whereas an interaction that understands and uses context is bi-directional; it is ultimately a real conversation. It’s important to highlight this because whether a conversation is singular or bi-directional with a client will directly impact the success of a customer journey.

Here’s an example of a how a typical chatbot would respond to a query, and one based on a real life Banking policy:

Customer: As the primary user I made a payment that has not debited my account, what do I need to do?

Chatbot: If you are a secondary user, the payment may have exceeded your single payment limit. In this case, your primary user (or secondary user with the ability to authorize a high enough single payment limit) may not have authorized it.

 If you are the primary user, or a secondary user who is sure the payment was within the limit, please call the Business Internet Banking helpdesk and we’ll look into this for you.”

We can immediately see the chatbot has given a “cover all the bases” response and the customer is being asked to reference their own context to extract a meaningful response. As a customer, this is neither compelling nor indicative that the chatbot is understanding the meaning of the question. This directly, and negatively, impacts the customer experience.

Virtual agents on the other hand are capable of holding more complex conversations with customers as a result of the science behind natural language processing. Unlike chatbots, virtual agents are able to contextualize human responses and define the underlying meaning of words, especially in terms of non-standard conversational language.

Through virtual agents’ advanced natural language capabilities, human responses or questions no longer have to be “perfectly” phrased to process a request—reducing the amount of mental effort that a customer needs to exert when interacting with an organization. Also defined as a heavy cognitive load, customers that exert too much mental effort in the beginning phase of an interaction tend to develop a negative outlook on their overall customer service experience.

Here’s how a virtual agent like Amelia would handle a customer query:

Customer: As the primary user I made a payment that has not debited my account, what do I need to do?

Virtual Agent: This can happen if you’ve exceeded your single payment limit for the day, and can often be caused if you’ve authorized someone else to make a payment on your behalf as a secondary user. Is this something you’ve authorized?

As you can see, the virtual agent has understood the context accordingly during the conversation and generated a meaningful dialogue to assist through bi-directional communication. It recognized the context (that you are the primary user), and then engaged in a unique line of questioning to help the customer with his/her request. It’s a much more direct, personalized approach to resolving a customer issue because the virtual agent does not overload the customer with information. It moves beyond simply responding with text and moves into a process that tries to help the customer by identifying common causes with a triage process typically undertaken by a real life agent.

Depending on the business area an organization is trying to optimize, chatbots and virtual agents can provide a significant boost to operational efficiency and productivity as long as they’re deployed within parameters that won’t exceed their functional limits. For first-time adopters, it’s critical to think carefully about what aspects of your business you may want to automate and decide what type of AI interaction is best suited for your needs—chatbots or virtual agents.

Ultimately both types of technology—although markedly different in functionality and capability—are ushering in an age of automation that will radically improve the way businesses across multiple industries approach their customer service operations. It’s an exciting journey for both the business and the consumer to be a part of.