Fortunately, the unrealistic expectations regarding how conversational AI would allow chatbots to be almost fully conversational have been tempered.
As the industry matures the right UX approach will emerge and in our view, this approach will be a blend between Natural Language Processing (NLP), scripted interactions and graphical UI within a chatbot.
This trend can be seen in Facebook’s partial shift towards making Facebook Messenger the universal mobile app platform through its emphasis on graphical widgets and in particular web views and chat extensions which allow developers to build app like interfaces within Messenger.
There are now an abundance of platforms that allow people to create scripted chatbots very easily, via visual tools such as Chatfuel, Octane.ai, Motion.ai. These tools are great at what they do and reasonably sophisticated chatbots can be created without any code.
These platforms have limitations however in terms of how easy it is to customize the bots (in terms of behaviour and custom UIs) or integrate with internal systems.
There is one area in the chatbot space that has exploded, and that is the use of broad, superficial AI. In this space NLP excels. The likes of Alexa, Siri and Google Home allows for a voice or text UI based on broad but superficial voice commands. These types of one off interactions make the best use of the tremendous advances made in NLP of late.
So what does this mean with regards to developing your chatbot for the mainstream? We believe you should be aware of the following trends in designing your chatbot UX.
- Chatbots will use fully customized graphical interfaces. Even if the NLP was perfect, the chat interface is not optimal for doing many types of tasks. On a pure text interface, it is difficult to get an overview of the options available, the interactions happen only in one direction, it takes many touches to issue an instruction, it’s difficult to revise a previous answer and difficult to keep track of the status of relevant information. There are many jobs therefore that are best done through a graphical interface. Web views are just the beginning of this trend.
- The chatbot UX will be truly flexible. Chatbots will allow the user to accomplish the same tasks or a series of tasks in multiple ways, with a combination of graphical widget, voice, and text. A user will be able to choose which interface suits the task at hand in the given context.
- Chatbots will offer cross platform continuity. A user could start a task on Slack, continue it on Messenger and complete it in a Web Chat. The thread will exist across platforms.
- Chatbots will become more job focused. There will be in general less emphasis on non-essential text and features. Content will still be important, but the idea that content alone can keep people interested in a scripted, mechanical conversation will be proved false.
- Chatbots will come to represent a task focused, personalized journey for the customer rather than a specific technology or approach.
What does this imply for chatbot development?
- As with any new technology, there will be more emphasis on the reusability and sharing of functionality as chatbot development matures. Sophisticated chatbots can only be built if there is a solid infrastructure of common functionality that can easily be built on.
- There will be more and more customization. The flexibility and customized UI of the chatbot will require more developer expertise. Coding and related professional practices such as source control, testing and deployment procedures will be central to the process of chatbot development.
- Specialization will happen in development roles. The roles of flow professionals (developers) and content professionals (user interface designers) will diverge and become more specialized.
- Specialization will happen for platforms. The bot will need to be customized to make the best use of the features of each chat platform, for example, Slack has different features to Messenger which has different features to Kik. The development tools will make it as easy as possible to use the best features of each platform without duplicating effort.
Some related trends for frameworks such as Botpress are as follows. Chatbot development is bound to become easier in future, allowing chatbot makers to spend more of their effort on the UX:
- To encourage reusability and sharing through a modular architecture but at the same time allow for total customization through code.
- To allow for and encourage the separation of and professional management of code and content
- To make it as easy and efficient as possible to customize a bot to use the best features of the chat platform in question, through graphical elements management.
- To offer professional administration tools such as multi-user management and role based security.
- To make it easier to use and switch between NLP platforms.
- To offer and encourage integrations with relevant third-party services.
- To build advanced human in the loop capabilities that support communication beyond just text i.e. that support communication using graphical widgets.
- To offer marketing tools that support segmentation and A/B testing among other things.
- To build in physical web capabilities to enable bots to perform local, geolocated services.
We will be watching how this space develops and will keep you posted regarding any new trends that start to emerge.