In order to present our vision of this collaboration, we partnered with SAP Conversational AI (formerly known as Recast.AI) develops a technology based on natural language processing (NLP) and provides it to developers through a collaborative platform to build bots. For companies, this startup specializes in the automation of customer support.
The growth of chatbots
Chatbots can be implemented in a wide variety of channels: Messenger, live chat, SMS, Viber, WeChat… Facebook Messenger is for the moment the main platform, with over 200,000 active chatbots. Messenger actively supports their development, by encouraging, for example, users to interact with chatbots via its “Discover” tab.
80% of businesses want to use a chatbot by 2020. According to this study, the number of chatbots will rapidly grow and reach nearly 25,000 by 2022. This growth is comparable to that of websites between 1996 and 2001.
“This momentum is a reliable fact”, says Jasmine Anteunis, co-founder of Recast.AI. “Since our foundation in late 2015, we have developed into a growing bot-centric ecosystem. Today, we have over 30,000 developers that created 60,000 bots, and 20+ large companies clients. The demand is here.”
These figures show that the use of this technology is just beginning, and that companies will gradually learn to improve it by implementing the appropriate models, combining humans and artificial intelligence.
The current usage of chatbots: a specific case
When launching a chatbot project, it is important to keep in mind its technological limitations. A bot will always be a computer program based on algorithms. It cannot provide an answer to everything. People using chatbots for the first time often expect similar results as in a human conversation and may be disappointed. Developing a microbot, specialized in a specific field, will mitigate this disappointment.
It has been demonstrated that the use of chatbots may be more appreciated by customers in certain situations: most prefer to deal with chatbots to obtain information about a product or an order status; while they prefer to interact with a human for a claim or making an appointment.
Hence, it is important to understand how we use bots today. A chatbot does not know what it has not been taught. Therefore, it is only efficient in its main use i.e. the one for which it was designed.
Creating a bot that does everything is a very complex task and it makes more sense to design an agent that can handle well a single subject.
The more precise the use, the more positive the experience will be for clients. From a technical point of view, the understanding of language will be more focused on a small area of activity; from the point of view of use, a bot that claims to handle a single topic manages better user expectations than a bot claiming to cover all topics.
An example scenario of a chatbot designed for a specific area
Ultimately, the user must be informed that they are talking to a bot in order to better handle their expectations: if the bot mimics a human personality, users will expect greater agility in their conversation than what the bot can handle, which may lead to many frustrations.
How bots can enhance humans
AI has inspired all sorts of more or less optimistic forecasts. Among these, the fear that artificial intelligence could replace the work performed by humans and thus kill jobs. The applications of chatbots, as well as their limitations, show that the key to ensuring the best customer experience is combining humans and chatbots.
Several leaders are currently defending this idea: according to Eric Schmidt, former CEO of Google, AI will create new jobs and give employees the opportunity to develop new skills. Barack Obama also believes that technology will contribute to an improvement in skills and living standard. Gartner confirms these positive consequences and predicts that by 2020, AI will help create 500,000 jobs.
Bots are not superheroes. Today, they perform well for their intended use, but they are not meant to replace humans, the only ones able to analyse complex situations. Furthermore, the human-bot complementarity can prove very interesting. A bot is able to handle a large volume of repetitive tasks, such as tracking orders or managing accounts, while humans focus on tasks with high added value such as pre-sales or after-sales service. Combining the strengths of each allows to provide a high-performance customer service at low cost, without drastically transforming existing organizations.
How does this collaboration work?
The human-chatbot collaboration is crucial to ensure an optimal customer experience. However, ensuring collaboration and a smooth transition between the two parties is a new challenge for companies.
Several scenarios can be adopted:
- An independent chatbot that will redirect to an agent when necessary
- A chatbot at the beginning of the conversation that collects information before redirecting to the appropriate agent
- Switching back and forth between a human and a chatbot during the conversation depending on the actions to be performed
These scenarios provide the user with quick answers, at any time and in any language, while offering the possibility to converse with a human agent, if needed, hence ensuring quality customer support on topics that the bot does not manage.
Faced with a multitude of digital channels (e-mail, messaging, chat, etc.) and the imminent arrival of channels developed by GAFA (Business Chat, WhatsApp for Business, Amazon Anytime) that could also host chatbots, centralizing management requests has become a crucial need. A customer interaction management platform allows agents to respond to messages from multiple sources through a single interface.
A customer interaction management platform at the centre of human-chatbot collaboration
By connecting to chatbots, this type of platform allows a handover between humans and chatbots according to rules defined by the company. Several scenarios that we presented previously can be set up quickly. This type of scenario ensures the best possible customer experience and avoids disappointing them.
These perspectives show the importance of the potential of chatbots for customer care. Advances in this technology and the continued growth of messaging applications will further increase their use in the coming years.
Increasingly advanced and reliable models based on chatbots are being put in place. These applications contribute to enhancing customer experience, with a seamless implementation and by providing instant answers 24/7. Practice shows that humans still play a crucial role alongside chatbots, and that their collaboration will be beneficial for businesses, thanks to happier customers, more satisfied agents and an evened workload during the day.
Originally published Mar 07, 2018, updated Sep 23, 2021