After three hours a chatbot on your own mobile: tangible result of a chatbot workshop organized by Pegamento. Although there is still work to be done on a good customer experience, developments are moving quickly. Companies that do not want to miss out in the future are advised to claim their 'invocations' in the short term. huh, what? Invocations?

 

There were about ten participants at the table, including solutions architects, change managers and customer contact specialists from different organisations. In most cases, there was already some experience with bots and new applications were being worked on. For example, for serving customers outside regular opening hours or setting up an onboarding app. Of course, the participants in the chatbot workshop were presented with a view if necessary.

AI can register call reasons very well

There is still a huge potential for automation possibilities. Our environment contains a lot of information; when you offer these (whether or not combined) to a learning system, you can extensively automate all kinds of standard processes. Think of generating a quote for exterior painting based on a photo of a house (where image recognition and geolocation are linked to cadastral data). Or arranging a repair appointment online and mobile directly with car damage. During the chatbot workshop – with mainly female participants! – Pegamento takes a look into the future.

For example, it is very obvious that not agents, but AI will soon record the call reasons in the CRM system. And it is quite likely that companies will soon have groups of bots that together support or execute all kinds of processes. For example, Airhelp, a company that helps airline passengers submit claims (delays, etc.), already works with four different bots. One analyzes the incoming cases, the other looks for new patterns in exceptions, and a third bot helps to retrieve missing documents. Number four supports the lawyer.

All opportunities for technicians with a customer feel

The developers of Pegamento are convinced that for empathetic technicians more and more opportunities lie in customer service. When developing dialogues, you must not only be able to think in a structured and logical way, but also be able to put yourself in the shoes of the customer, understand customer service processes and have a good sense of language. And being able to work well with conversational platforms such as Google's DialogFlow: online applications that are being developed at a rapid pace with multiple releases per month.

A good chatbot can classify the intent (intention of the customer), which is then necessary to take appropriate action. During the workshop, the participants were guided through the development process with DialogFlow piloted away. Working with such a platform requires, on the one hand, knowledge about how your customer asks his/her questions; on the other hand, a feeling for language and technology is also indispensable. The latter helps in properly assessing the bot's information needs: when will there be a substantive answer to your question instead of 'I don't know'?

Dialogflow already contains an enormous amount of basic knowledge and skills. For example, Google can link a general indication such as 'tomorrow' to a calendar and understanding 'Friday afternoon in a week' is also no problem. Dialogflow also quickly learns to deal with ambiguous information in sentences.

New platform and channel choices for customer service

The conversations you design in Google Dialogflow can be 'deployed' on multiple platforms at the touch of a button: think of Facebook Messenger, Slack, Telegram, Kik, Skype or Line. Also, the bot processes are primarily text-based, making them suitable not only for speech, but also for texting or email-based customer contact. In addition, Google is not alone (speech) platform. Cortana, Siri and Alexa are also participating and it is still too early to say that Google is definitely taking a leading role.

The outcome of the chatbot workshop was extremely practical: the start of a working solution where you call Google on your phone (Hey Google,…) after which your own bot takes over the conversation and can take two to three steps. DialogFlow is somewhat complex for laymen, but anyone with a bit of 'tech savvy' should be able to get a long way. During development, you can test every realized step yourself in a simulation environment and there is a lot information for developers readily available.

Claim your nickname

In addition to a rudimentary, but working chatbot, Pegamento also provided a number of advice. Companies must claim their invocations in a timely manner. An invocation is the phrase with which you call Google Home Assistant ('hey Google) and also the unique phrase with which you let the Assistant switch to your own chatbot environment. For Rabobank this is (after 'hey Google') 'talk to Rabobank' – after which you will be 'transferred' from Google Assistant to the Rabo chatbot. Google has already reserved the most important brand and company names for the owners of those names. Names such as bol.com and KLM can only be claimed if you can prove that you are the owner. More generic names – including those of services, for example – can still be used for naming invocations, so that you as a company can set up your own unique services. Think of 'wine delivery' or combinations such as 'Auping bedtime story'.

In addition, companies must think about the translation of their identity to the voice channel: how will our manners with customers become? How do you present yourself, what relevant services do you offer and how do you become easily findable and distinctive? Enough new challenges for digital customer service.

 

Pegamento will organize similar workshops more often in the near future, where customer contact professionals and developers can get started developing chatbot dialogues themselves. Keep the calendar (also in the Customercontact.nl newsletter) for new sessions. 

Text and image: Erik Bouwer

Featured, Technology Glue