A chatbot understands our language and responds as a human being if a conversation designer, i.e. the architect of virtual conversations has set the conversation agent below.
At the basis of any interaction or conversation there is the ability to understand without ambiguity and to respond coherently to one’s interlocutor and in this sense communication between humans and chatbots makes no difference.
A chatbot understands messages through NLU (natural language understanding) technologies that analyze and interpret natural language using complex machine learning algorithms.
Conversation designers are specialized in the construction of conversational flows and must be as natural and fluid as possible.
As computational linguists, they are involved in algorithm training, powered by sets of possible questions, technically called…User Says, which the machine uses as examples to be followed to provide the correct answers to the user.
Conversation designers are essentially responsible for managing the NLU component of a chatbot, i.e. processing the inputs that arrive at the system to allow the chatbot to understand the requests of users, and the design of the conversational flows, i.e.
The conversational flow is collected in the chatbot database, which draws on that information to select the most relevant and relevant response, so as to resolve the alert of the person who requests it in a short time.
These answers, besides trying to be as natural as possible at the conversational level, must also be easily available through guided paths or spontaneous questions, so within many chatbots are integrated automatic tips or buttons to guide the user to find the solution more easily
To do all this, however, conversational interfaces need a voice, personality and context awareness similar to human ones.
Here are the speakers, specialized in identifying these characteristics in order to create a customized, engaging and unique Customer Experience for every type of request that the chatbot will have to respond to.
Conversation Design Heres is a startup for the development of multichannel and multi-language conversational agents to automate business processes and improve customer experience. Founded in 2017, it has 25 active bots for customers such as Prénatal Group, Motoabresole, Admenta Italy, Easycoop, Shimano, Ianiri Informatica, Istituto Italiano di Tecnologia, Heres’ strength is his Conversation Designer, specialists in conversations that define the conversational experience, making the bot proactive and empathetic, designing every aspect of the Virtual Agent and customizing it on the basis of the interaction processes and the Heres has turnover 360 thousand euros in 2020 and plans to invoice 600 thousand euros at the end 2021.
How a Conversation Designer Works
Let’s see how a conversation designer works in team, taking inspiration from Heres.ai teams in developing a chatbot: what are the responsibilities he must take to build a virtual conversation that meets the expectations of users.
What the customer needs
The first competence in project development is to descend into the specific context of the customer to understand the needs and find the most suitable conversational solution. In the early stages the approach used is very similar to a user experience project: the client is involved in an assessment interview in order to identify the criticalities of the current assistance, the objectives that are to be achieved with the insertion of a chatbot From the briefing, the team analyses the emerging needs, the knowledge base already present on the customer’s touch points and all the materials useful for the construction of the communication flow. All this initial work is precisely what distinguishes the enterprise conversational agents from the so-called “Do it your self”, that is chatbot…no code… where the customer is required a great effort to create questions and answers without
Understanding the user
What does it mean to give a voice to a chatbot? It means to care for the communicative component, that is to provide the user with the answer he seeks, involving him in a pleasant and smooth dialogue. The conversation designer team identifies three steps in the construction of a successful conversational agent: understand what the user asks, guide him towards the right response and customize the final experience. As in any respectful conversation, the first principle to be respected is to know how to listen and, above all, to understand what you are talking about.
Even in virtual systems, understanding the needs of users is of primary importance if you want your chatbot to be successful. But identifying the correct intent of the user is not a simple task. This is because, although NLU technologies can carry out a semantic, syntactic, phonetic and morphological analysis of written and oral statements, the transition from understanding to pragmatic remains a real challenge.
Since human languages are formally complex and highly context-dependent, a request can be expressed in so many different and ambiguous ways that it is often quite unlikely to map them all, even for the human being.
Conversation designers, however, try to find infinitely replicable similarities (patterns) in user requests and to feed NLU algorithms with examples that can help them associate a precise request with a given answer. To do this, different actions may be needed by the conversation designer, such as focusing more on one or another of the language analysis levels, using a proprietary NLU platform or, as in the case of Heres.ai, relying on one or more of the following:
Driving the user
The second phase of a conversation planning is the design of the flows that will guide the user towards all the answers in the knowledge base.
The conversation designer has the task of building conversation through two modes: creating simple answers or multi-step, parametric, deep-response workflows to the user’s spontaneous questions and guiding them along specific paths with automatic tips and multiple-choice buttons. The combination of these methods in the same conversation ensures a better conversational experience.
Here, at this stage, the conversation designers of the Bolognese start up plan structured tree paths based on a careful analysis of the domain and specific topics managed by the chatbot.
They can also customize these responses on the basis of external factors, such as the site page where you are on or the time of the day or year, and other factors such as user’s registry and habits, language, age, number of orders or purchases This data may be requested when you are talking to a virtual system or called up by managements used for marketing automation.
What if the chatbot does not immediately understand the intent of the user? Conversation designers have also thought about this by providing routing paths for custom conversations to guide the user by climbing to a second level of human assistance.
Customize chatbots and final experience
The tone of voice of a conversational agent naturally depends largely on the brand involved and the target of users to whom it is addressed. The ability to equip the chatbot with a personality close to that of the brand is fundamental: the chatbot is in fact at the forefront of the relationship with the brand’s customers by presenting itself as a top-level assistant on the dedicated touch points of the Customer.
Writing chatbot conversations doesn’t just mean that grammar is correct, proper tone and accurate lexical choice, but also playing with words, cultural standards and shared knowledge, so that the chatbot’s answers look like
The truth is that there is no perfect UX conversational, but only flexible conversation designers who can shape a chatbot to the needs of customers and end users