Virtual assistants are gathering attention in the business world, and it’s easy to see why. They can answer FAQs and speed up customer processes, they have immense value in gathering marketing leads, they’re cost effective, and - among many other things - they’re pretty cool.
Ezra Chatbot Accelerator Version 1.5 is here and believe us, this hype is real.
Ezra is more than a bot framework, it allows you to create your own business chatbot with a natural conversation flow, minus the coding pains. Ezra is extremely customisable, but with a user-friendly configuration interface: you don’t need to be a developer to build this bot.
Ezra navigates your information universe so that customers and internal staff can easily have their questions answered irrespective of which system the information is stored in.
Natural language understanding is essential in a chatbot that can conduct a sensible conversation, and Ezra utilises AI technology to understand the intent of a question, and then return the relevant answer from its Knowledge Base. You can help Ezra constantly learn from its past conversations by using the integrated Power BI model for immediate insight.
Here’s a sneak peek into how you could build your own Ezra v1.5
The dictionary is where Ezra’s terms and phrases are entered. These words will be used as tags for answers that you save into the Knowledge Base. This allows you to personalise Ezra for your business purpose, and decide how Ezra will interpret questions. You can set the function of each dictionary word with word types: verb, query, noun, location, person. Verbs and queries inform how Ezra will phrase its response, based on the Answer Grammar Matrix (more on that later).
This is basically Ezra’s brain, where Ezra will pull answers from, depending on how you tag your Knowledge Base inputs. You can also tell Ezra to look for certain answers in external databases. It’s very easy to tell Ezra exactly what to say. You can enter a SQL statement if you need information to be selected from a SQL database, you can send users straight to a URL link, configure Ezra to answer with a photo, and much more. There’s a snazzy scoring algorithm that Ezra uses to make sure the best possible answer is returned. See, we told you it was cool.
Ezra really begins to show off its skills with the Action Flows – these allow you to manage conversations when Ezra is asking the questions. This is used for capturing information from the user or getting the user to re-phrase the question if Ezra doesn’t understand. You can get Ezra to facilitate filling in customer forms, or surveys, or even escalate the conversation to a human customer service operator. Action Flows are even easier to create in version 1.5, with the new and improved Action Flow builder, that can pan and zoom and let you build as complex a flow as you need.
Answer Grammar Matrix
The matrix formats how an answer is returned, in respect to how the question is put forward. Essentially, it helps Ezra sound more conversational, and more human. Each Answer Grammar is defined by a verb type, a query type, and whether Ezra has an answer to return. In the example below, you will see that verb and query types from the dictionary are plotted. Where they intercept is where Ezra will craft an answer type (the answer grammar) to respond to the question.
If all else fails, Ezra will turn to several web search engines to ensure an answer is given. This means that you can ask Ezra about almost anything – “what’s the time in London?” or “How old is Donald Trump?” and get an accurate answer. Ezra will attribute the search service it pulled the answer from.
Read more about Ezra as a business solution, or test Ezra out in the chat window below!