- 16.04.2021 04:00 pm
- 16.04.2021 02:15 pm
- 15.04.2021 06:30 pm
- 15.04.2021 03:45 pm
- 14.04.2021 04:45 pm
- 14.04.2021 04:30 pm
- 14.04.2021 04:15 pm
- 14.04.2021 03:45 pm
- 14.04.2021 08:00 am
- 13.04.2021 06:30 pm
- 13.04.2021 05:30 pm
- 13.04.2021 10:15 am
Chatbots are gaining in popularity in a number of industries as an important customer service tool, with financial services and insurance particularly keen to roll them out: Crédit Agricole Assurance has Marc, and Bank of America recently announced it was introducing Erica. Barclays, Société Générale, USAA, BBVA, and Capital One have all also begun investigating the technology.
The rise of chatbots is being driven by several converging trends: the popularity of messaging apps, the explosion of the app ecosystem, advancements in artificial intelligence (AI) and cognitive technologies, conversational user interfaces and a wider reach of automation. Their adoption is accelerating so quickly that Oracle believes that 80% of brands will be using them by 2020. But will the current hype be sustainable over time without a stronger business rationale and better short-term results?
We live in an age of instant gratification, and this certainly applies to exchange of information – the core mission of financial services. So why are customers confronted with long wait-times on hold, being transferred department-to-department, or having to wait through a list of phone prompts? In the context of chatbots, it is actually not about “the robot” at all, it is all about how easy the end-user finds it to use, and simply whether it works or not. To get it right, businesses should start preparing for the coming bot age now if they have not begun to already. This means peeking into the future and designing bots to respond to today’s customer needs, such as personalisation, context, meaning, first contact resolution, management, as well as bot-human interaction and interface design.
Here are four areas chatbots will evolve.
For chatbots to be effective, they need to become far more specialised in topics and tasks, and have the ability to personalise interactions. As time goes by, we will begin to see this happen. Very soon we will see expert chatbots that specialise in providing information about different banking solutions, while there will be some like x.ai’s Amy, Apple’s Siri or Microsoft’s Cortana that are experts in making calls and scheduling meetings, or helping to orchestrate process steps. For example, your close of escrow got delayed due to unforeseen disclosures from the seller – was the bank notified not to fund the mortgage loan? In-the-moment examples like these will make chatbots more utilitarian and dependable.
On the flip side, users will also then need to understand what the chatbots does, specialises in, excels in and – most importantly – where it has limitations. This leads to one of the most crucial design decisions: the history of continuity and personal connection. Consider this element as a “tuning fork” of sorts that brings together and harmonises all interactions a person may have on a given subject.
If the user were to stray from a central line of main dialogue, for example, from Siri to Facebook Messenger, a chatbot will need the history and context of other discussions with people, places, and things in order to provide continuity and personal connection. In turn, this will dictate how much personalisation can be brought into the interaction itself. For instance, can the system remember user profiles, previous interactions, the interactions of other users in the system, the current context and the situational bigger picture? Chatbot creators will then need to design them so that they can access this information using a multitude of systems and derive meaning from that information, all while keeping the central “plot line” of context intact.
One of the things that makes most present-day browsers so useful is their ability to answer questions at almost the speed of the user’s thoughts – sometimes faster. The experience of a good chatbot interaction is not judged only on its capacity to answer a question correctly but also the speed at which such a response is provided. In the bot world, solving a problem after a first contact with a customer will become a key performance metric.
Chatbots that can provide basic solutions in the first instance without the need to paraphrase or explain the problem in greater detail will be the most useful and, by extension, the most popular.
The concept of a superbot is not yet well known but will be a significant element in the future of bots. Indeed, as bots become more specialised and popular, they will proliferate. For many companies, managing them could become as overwhelming and complex as managing apps is today. The solution could come in the form of a superbot.
A superbot, or “bot of bots,” would act as a personal assistant, getting things done on behalf of the user. That would mean calling other bots to complete tasks such as scheduling meetings, dialing conference call numbers or redirecting the customer to the appropriate page to make a claim. The superbot would know which bot to call for a particular task and instruct that bot to provide feedback to the user, therefore being faster and more efficient. Some platforms already use “global managers”, automated robots that orchestrate workflow, and delegate which process transactions should be worked on by myriad other robots.
Many of us will have seen an example of a gimmicky, humanoid “greeter robot” deployed in your high-street bank branch but the chances are, it fell short on actual needs-based problem solving for the customer. Chatbots, to the rescue – customers actually want solutions to process common choke-points in the gaps between information flows. Most of today’s technology exploration focuses on enhancing features and improving functionality to enable chatbots to mimic human responses, engaging in a more natural, intelligent conversation with users. Despite the merits of this work, the continued success of chatbots will not wholly depend on their ability to conduct a natural conversation but on the accuracy of their responses to customers’ questions at the moment-of-truth: when the tax bill is due, when the overdraft charge kicks in or when the mortgage documents are being finalised.
Humans can sense when they are interacting with a machine, and any attempt to make it appear more human rather than intelligent may end up triggering negative emotional responses in humans— this phenomenon has been called “the uncanny valley” by a Japanese roboticist in the 70s. That is why some novelties robots are merely a distracting detour on the road to real breakthroughs in applying automation that matters to the financial services sector for real and lasting results.
Chatbots will be the vanguard of these efforts, and success will hinge upon their ability to become useful, maybe even indispensable, to human beings. Automation has its limits — and there are some things that robots just cannot do. That is where a blended model of automation augmenting people in their daily lives, conversations, and information requirements can provide extraordinary outcomes. By connecting conversations with meaning, context and intelligence, and providing people with relevant information in real-time and after absences, chatbots will provide as higher quality service and outcomes.
For companies in financial services, in addition to other industries, it requires striking a balance between speed, specialisation, and personalisation provided by chatbots and the ability to cater to human sensibilities and expectations. After all, the main goal is to support users and to make their lives easier.