Artificial Intelligence (AI) model
An AI model is a program that has been trained on a set of data to recognize patterns. It can be continuously improved through training. The more data, the more patterns, the more accurate your AI. You can think of your AI model as your virtual agent’s “brain”.
Bot handled rate / automation rate / resolution rate
The amount of fully automated, resolved, or “handled” conversations, out of all bot conversations in a chat.
A chat conversation is the sum of all messages exchanged between a customer service agent and a customer.
An automated dialogue logic that a virtual agent can follow to solve requests. Conversation design is interactive, meaning the virtual agent can ask questions to then give personalized or highly specific replies based on your customers’ answers.
The percentage of conversations that were not escalated to an agent, including conversations where the virtual agent did not understand the customer, or in which the customer abandoned the chat.
The percentage of conversations that were successfully escalated to an agent, out of all conversations with your virtual agent.
Expressions are customer messages that have been matched to what’s called an intent by your AI.
Examples of expressions include “Where is my order”, or “When will my package arrive?”
An intent is a group of expressions that all have the same purpose: They’re a question your customer wants to know the answer to. A good AI model can group expressions and create intent categories instantly and automatically - so your agents don’t have to!
An example of an intent would be “Order status”.
There are two types of integrations. The first is between your virtual agent solution and your CRM. The second is between your virtual agent solution and your backend (servers, applications, and databases). Integrations are a two-way street: Your virtual agent can pull information from your CRM or back office software, as well as update that information based on input from the customer.
A message is a part of a chat or email conversation. Several messages can make up one conversation. From “Good morning” to “I’d like to book a trip to Florida” – everything your customers or agents say to each other is considered a message.
Natural Language Processing (NLP) and Natural Language Understanding (NLU)
Natural language processing (NLP) is a technology that helps computers read and understand natural human language. Natural language understanding (NLU) is a subcategory of NLP. Using NLU, an AI model can learn a language using examples. It can understand and group a range of sentence structures and spellings, including synonyms and typos. Automating with NLU means never restricting communication to a list of predefined keywords, reducing frustration and improving your automation rate through the AI’s high accuracy. This is useful both for automated chat and email ticketing.
A ticket is a documentation of any sort of interaction with your customer. Some customer service providers also refer to a ticket as a “case”. While it’s possible for a chat interaction to turn into a ticket, you’ll usually use the term when talking about email or messaging apps.
NLU technology can help your virtual agent triage and answer your emails and messages, too. The first step is categorizing intents by applying tags or labels. Next, your tagged emails can be routed to the correct agent or department. With the added help of CRM integrations, you can then update customer information, or trigger automated replies.
(Intelligent) virtual agent
An intelligent virtual agent, also known as an intelligent virtual assistant (IVA), is not just a chatbot. It’s more like the agent behind the chat widget that you see in your browser. Its tasks can range from automatically fetching, routing, and labeling information to communicating with you directly via chat, messaging, or email.
Knowledge is power; Let’s wield it to get you started with automation.