Getting Started

Welcome to Ezra’s support guide.

This guide covers Ezra’s features, helping you through the next steps for getting your bot up and running after installation. You don’t need to use code to build Ezra. Ezra can be managed and trained by business people, without requiring specialist technical skills.

Scroll down for a snapshot of what you need to bring Ezra to life, or delve deeper into the step-by-step functionality of Ezra with the navigation menu on the left.

For any extra assistance, or to provide us with some feedback, send an email to info@perspectiveilm.com

To see Ezra in action, hop on over to our webpage: perspectiveilm.com/Ezra

What can Ezra do?

The Ezra Chat Bot Accelerator can connect to all corporate and external systems and start helping businesses anticipate the needs of customers and internal staff. Ezra can be configured by business users to display the exact information needed, without relying on IT for support.

As a snapshot, Ezra is capable of:

Generating and capturing leads by guiding customers to quicker answers through automated processes, eliminating the need to sit on hold or navigate through an entire website. Ezra can clearly offer calls-to-action based on the nature of conversation, and capture leads at the point of contact. This can be done by connecting Ezra to pull answers from an existing database, or by entering your information into its Knowledge Base manually.

Natural Language Understanding, providing a friendly conversational format. Ezra can understand the intent of a question and find an answer using tags in its Knowledge Base. Ezra’s answers can be easily refined with continuous feedback.

Giving meaningful insight via an integrated Power BI model, which produces instant data on the most commonly asked questions. Ezra then allows a business user to tailor answers according to this conversation history.

Where do I begin?

Some of the best, and simplest, examples of using Ezra are: for Staff Directories, Customer Service, and Information Redirection - scenarios that have a high-volume of interactions/traffic and require only simple responses. Ezra can field any number of related questions at any time, returning information in a friendly manner from databases or online, providing customers and staff an efficient self-service solution and freeing up resources.

More advanced scenarios can even have Ezra handing over Customer Service conversations to real people, complete with analytics on the caller’s sentiment and what products/options would be the best fit to ensure the conversation is a success.

Ezra can be very versatile, an advanced digital assistant that can utilise other advanced services or custom programs to drive the solution you want. Feel free to reach out to us and we can discuss how to make your ideas a reality!

Getting Started

Dictionary

The Dictionary is often the place you start when first building Ezra. When adding to your Dictionary, have a clear idea of what kind of things Ezra is going to be asked about, and the information you want Ezra to reply with.

Dictionary words are especially important because they are the tags Ezra needs to match user words or statements to - also called user utterances. Dictionary words or phrases need to be given a type: Verb, Query, Noun, Person or Location.

1

Knowledge Base

Ezra’s Knowledge Base is exactly that - a store of information that Ezra can retrieve answers from. Each entry is tagged with one or more words from the Dictionary, and the best answer is selected according to a system of tag scoring. This will be explained further in the Knowledge Base section.

2

Action Flows

Ezra has two modes. One is the usual free text Q and A (the user types a question and gets a response) and the other is a wizard based workflow called Action Flows. This second Action Flow mode can feel more like a dialogue, as Ezra is the one asking the user questions. Using Action Flows, the business can configure the next thing Ezra will say, according to the customer’s responses. Action Flows can look complex in their mapped form, but the user interface allows for a straight forward set up.

3

Answer Grammar Matrix

The Answer Grammar Matrix formats how Ezra answers a question, depending on how the question is put forward. It is based on verb type, query type and whether or not Ezra has an answer to give. The Usage Category you assign to your Dictionary verbs and queries helps to inform the Answer Grammar Matrix.

matrix

Global Variables

Global Variables are assigned values that persist within Ezra through all interactions with the user. This feature will be expanded in upcoming updates.

Failover

If Ezra doesn’t know the answer to a question - i.e. it is not in the Knowledge Base - a response can still be found using Ezra’s Failover feature.

Ezra’s failover feature utilises Bing, Wolfram Alpha and a selection of other third party services to search for nouns from the user’s utterance. This means that Ezra can be asked unrelated questions like “what is the time in London?” or “what is the highest building in NYC?” and still return the user with an answer. Failover can be configured or disabled in Failover Configuration in your left navigation menu.

Here’s an example of failover:

failover 1

failover2

For a more detailed example of Ezra’s components, please explore the navigation bar to your left. Happy bot building!

Don’t hesitate to contact us if you require more assistance.