Relive our recent webinar and learn what it takes to build an intelligent chatbot, based on first-hand experience.
In our recent webinar, Designing a Conversational Chatbot Experience: Tales from the Trenches, Hristo Borisov shared best practices for creating modern conversational chatbot experiences. These insights were based on his first-hand experience of conceptualizing and building Progress Kinvey Chat, the artificial intelligence-driven platform for creating and deploying chatbots.
If you’ve tried to build a chatbot before, you’ll likely agree that achieving quality human-computer interaction poses challenges. Beyond natural language understanding and the personality of a chatbot, developers need to account for unexpected scenarios, ambiguities and other normally occurring human behaviors during conversation.
If you missed the webinar, worry not. You can watch the entire recording below:
Since we believe you can learn a lot from other people’s questions, we thought we’d share a few of the more popular questions that came out of the live event with you.
Q: What kind of experiences are suitable for a chatbot? Would a troubleshooting/helpdesk type of experience be suitable for a chatbot?
A: The current state of chatbots is that they can handle relatively simple tasks with well-defined steps. As a developer, there are things you can do to help improve the experience and engage in deeper dialog.
You will need to design an intelligent conversation based on the natural language processing intents and entities. You have to develop an algorithm for each conversation and a simple navigation. You can use simple decision trees, state workflows, a slot-based algorithm or some advanced deep learning algorithms to control the conversation. The more intelligent you try to make these bots, the less trivial it will be to implement them.
To enhance the experience, you can also implement a conversational UI that contributes holistically to the natural flow of the conversation. With the proper UI elements, the developer can provide visual guidance to the user that helps him/her navigate the conversation more effectively. Not only does this provide a better overall experience for the end user, but it means a more productive development cycle for the developer when it comes to NLP and conversational intelligence.
Q: Is it better to use a guided conversation or AI?
A: Neither is better; you can and should use both. We recommend you use AI for natural language processing to understand what the user is asking about and then guide them to a resolution with guided conversation.
Q: We often have the problem that, when the language model grows too much, the detection of the intents gets worse.
A: This is common today. The understanding should be monitored. We are currently devising some learning strategies that will allow the model to get improved automatically and we have some analytics reports that enables chatbot developers to see where the NLP gets worse so they can act on it.
Q: Is there anything I should consider regarding the storage of my chatbot data? I am thinking about GDPR specifically and the desire to keep data contained in certain countries or sets of countries.
A: You can definitely be deliberate about storing your chatbot’s data in specific countries. Our Kinvey Chat product will be GDPR compliant, and we are taking measures to ensure that a chatbot developer has all the necessary resources to make his chatbot GDPR compliant as well.
Want to Learn More?
If you can’t tell, chatbots are fast becoming a favorite topic of mine and many of the folks here at Progress. For more information about chatbots in general, check out “The Anatomy of a Chatbot” blog post right here on the Telerik blogs.