Let’s face it, the world of chatbots is filled with jargon and acronyms like AI, ML and NLP. We’ve put together this chatbot terminology guide to help you understand what you’re reading.
When you read about chatbots, it can be easy to feel overwhelmed by the sheer variety of technical terms and acronyms that surround the technology.
Sure, you probably have a general understanding of what chatbots do and perhaps even know the programming basics behind them. But suppose you need to look for a chatbot provider or have an informed conversation about artificial intelligence (AI) and machine learning. In that case, you need to beef up your chatbot vocabulary.
We’ve put together this chatbot terminology guide to save you from having to Google every developer-centric term and technical acronym you come across. This glossary will cover the basic chatbot terms you need to know and will be regularly updated with new entries.
First things first: a chatbot is a computer program that imitates and processes human conversations on websites, apps, and messaging platforms like Facebook Messenger and WhatsApp. They automatically process questions and provide relevant answers without human intervention.
While most people think of chatbots as being limited to written conversation, more advanced software can handle spoken language. These bots are sometimes referred to as voicebots.
Chatbots hold tremendous promise for customer service and stakeholder relations, with Mordor Intelligence projecting the chatbot market to reach $102.29 billion in value by 2026.
An Application Programming Interface (API) is a piece of software that allows two separate applications to interact. Think of it as a bridge that allows two apps to integrate.For example, the WhatsApp Business API allows chatbot makers to integrate their live chat solutions with WhatsApp, allowing brands like BMW to deploy a chatbot on the messaging platform.
3. Artificial intelligence (AI)
At its simplest, artificial intelligence (AI) is the simulation of human intelligence by machines, particularly computer systems.
In chatbot systems, AI enables chatbots to combine rule-based systems (i.e., predefined answers to specific questions) with context — the software analyses the context of conversations and compares them with historical data to provide relevant answers.
In other words, AI-powered chatbots perform messaging tasks that would typically require human intelligence for:
- Decision making
- Speech recognition
- Language translation
4. Chat widgets
Chat widgets are customisable chat windows that you can instantly deploy to your website. The chat widget is the primary interface linking your website’s visitors and your chatbot. Here’s an example of a chat widget on the Futr homepage.
5. Conversational AI
Conversational AI is the set of AI-based technologies that enable automated messaging and speech-enabled software to provide human-like conversations between computers and humans.
Conversational AI-powered chatbots use natural language processing and machine learning (both branches of AI) to understand customers and carry natural, human-like conversations.
6. Digital channel or conversational channel
Throughout our site, you’ll see us mention “deploying live chat and chatbots to your digital channels”.
These digital channels or conversational channels are places where we can launch your chat platform to interact with your customers or stakeholders. This could be anything from your website, app, Facebook Messenger, WhatsApp, Telegram or Slack, among others.
7. Machine learning
Machine learning (ML) is a technology under AI that uses algorithms (a set of rules or procedures for solving a problem) to identify and remember patterns in data — hence the word “learning”.
In the context of chatbots, machine learning allows bots to understand patterns in human language and better understand the context and intent of each keyword and query. The more data the chatbot parses, the more capable it becomes at conversing with humans.
8. Natural language processing (NLP)
Natural language processing (NLP) is another technology under AI, focused on understanding text and spoken words the way humans do. Together with machine learning, NLP combines rule-based models of human language with historical data to understand the intent and context of certain words and statements.
NLP also allows chatbots to translate text and speech from one language to another, as seen in platforms like Google Translate.
9. Natural language understanding (NLU)
Natural language understanding (NLU) is a subfield of NLP that seeks to understand human language. The difference between NLP and NLU boils down to objectives:
- NLP seeks to facilitate human-like communication between humans and computer systems.
- NLU focuses on a computer system’s ability to understand the context of human language. It seeks to decipher and rearrange unstructured language data so machines can understand them.
Think of it this way: before a chatbot can process a query and respond to it with a relevant answer, it must first use NLU to understand the query’s specific attributes.
10. Self-service or self-serve
Self-service/self-serve is a user experience (UX) feature that allows customers to complete an action or task without any assistance from your customer service agents.
Think of a petrol or charging station for cars, for example. These stations provide instructions on filling your tank or charging your electric vehicle — a relatively simple and pain-free process.
The same principle applies in the world of live chat and chatbots. By providing a list of your services and answers to frequently asked questions within the chat, customers can help themselves without speaking to your live agents.
Here’s an example of our chat widget signposting users to two service options: speaking to a Futr representative or booking a demo of our platform.
Self-service optimises your customer experience by giving users what they want when they need it. As counterintuitive as it sounds, many of your customers don’t necessarily want to speak to your company representatives. In fact, 67% of customers prefer using self-service options over speaking with a live agent.
11. Sentiment analysis
Sentiment analysis is a field under computer science that uses machine learning and NLP to understand the tone, intent and context of a text-based message or spoken language. It allows a chatbot to detect a user’s mood, largely by deciphering clues in their sentence structure and the words they use, connecting them to emotions like:
Sentiment analysis also enables chatbots to escalate complex conversations and queries by dissatisfied customers to live agents. If the chatbot detects words like “I’m tired,” “this is the nth time” and “I need…now”, it can understand the intent of the message, sense its urgency and immediately forward the conversation to your customer service team.
Sentiment analysis also plays a vital role in getting an overview of your customers’ overall mood based on common queries and the type of words they frequently use.
12. Software integrations
Software integration is the process of connecting two or more applications to work alongside each other. For example, with integration on Facebook Messenger, we can deploy your chatbot and live chat solution to your official Facebook page.
A more advanced example of software integration is enabling digital payments through chat. With Futr, for example, Stripe or PayPal integration allows users to pay for products and services without leaving the chat.
You can explore our list of 200+ live chat and chatbot integrations.
These are the platforms where you can expect your chatbot conversations to happen. There are plenty of communication channels available. Examples include Skype, Facebook Messenger, SMS, Signal and WhatsApp. When choosing a chatbot, one of the most important factors to consider is your knowledge base, but coming a close second is where to deploy your chatbots.
Clients, users and customers may expect you to use specific communication channels, so you need to ensure your chatbot is not only capable on those channels but excels on them. As chatbots become more mainstream, they are now capable of being deployed on multiple channels at once.
A very common word to come across when investigating chatbots, an entity is considered a defined variable that will occur in a conversation with a chatbot. That variable will be a word that lets your chatbot understand what the user wants. For example, if you run a retail outlet selling shoes, and a customer requests information about buying a blue pair of shoes, the words “blue” and “shoes” are the entities.
Understanding entities means that a chatbot can then understand the user’s needs and direct them immediately to where they need to go. For a social housing organisation, it might be common to get a chatbot query such as “I want to pay my rent.” The entities here would be “pay” and “rent”, so the chatbot would be able to immediately link to a payment portal.
15. Knowledge base
A knowledge base is a data repository intended to act as a self-service helpdesk for clients. They will take a variety of forms, from FAQ pages to troubleshooting guides and how-to articles. It’s where clients will go (or discover on search engines) to get fast answers to queries.
To make chatbots more intelligent, you will integrate them into your knowledge base. As clients or customers ask questions, the chatbot can scan through thousands of keywords and entities on your knowledge base, find the relevant information, and deliver it quickly.
To create conversational communication, chatbots need information, and the more information they have, the more they can learn. That’s where a knowledge base comes in. The fact that they can deliver that knowledge base information in multiple languages only adds to their value.
This will be a piece of information provided by the client or customer that the chatbot can remember and refer to throughout the conversation. Attributes are, essentially, small packets of information. The information that makes up an attribute might include a client’s name, an email address, or a phone number. As the chatbot carries on a conversation, it will collect attributes so that its responses can be more personalised.
Chatbots don’t replace humans; they complement them. In many conversations, chatbots will have to transfer a user to a human representative of the business. This will often happen when a user has a query or a problem that goes beyond the capabilities of the chatbot.
That’s why it’s important to train chatbots to recognise when escalation to a human is necessary. Providing a clear path so that chatbot conversations can be escalated is one of the 10 chatbot best practices.
There will be many occasions where the best outcome possible for an interaction is through the human touch. The key times for transfer to occur include:
Overly complex queries
While chatbots are very, very smart, they don’t have unlimited power. They will come across a query or task they aren’t capable of responding to adequately. Occasionally, this is due to the way a user has phrased a question, which would require a level of creative thinking the chatbot isn’t capable of.
Smart chatbots know their limitations and know when a user needs human interaction. They then pass that user onto a human agent, who won’t have to ask the user to repeat any conversation or questions because they will have a full transcript of the conversation in front of them.
In some cases, a back and forth conversation with a chatbot is the last thing a user needs. There are simple scenarios where the situation is critical and a human response is required immediately. The police forces that use chatbots are a good example of recognising when urgency is critical.
In emergencies, where a rapid solution is required, the chatbots recognise that intent and pass the user onto a human. The chatbot can then return to answering the less urgent queries.
While chatbots can easily close sales for a business and even upsell, not every company wants their chatbots to do so. Brands in the B2B sector are a good example of that. Chatbots will react to an evolving conversation in real-time, so when it recognises an impending conversion, the chatbot can be trained to transfer to a human as needed. The lack of wait time improves the chances of making a sale.
18. Typing delay
One of the most important advantages of chatbots is that they are capable of replying to a query instantly. The problem with those instant responses, especially in complex conversations, is that they stop feeling conversational and natural. That’s where typing delay comes in. Using a scrollbar, companies can initiate a delayed response, so there is a more natural interval, and therefore flow, between responses.
That delay can be anything from 0.1 seconds to 10 seconds, and during that designated delay, the user will see an indication that a response is being typed. This allows for previous messages to be more thoroughly read and understood, improving the experience of the user. So while a bot can reply to a query within milliseconds, it’s not always best to do so from a user perspective.
When in doubt, get in touch with the Futr team
This chatbot terminology guide covers the basic terms you’ll encounter in the world of chatbots. If you need to learn more about the technical nitty-gritty of bots and AI, don’t hesitate to contact the Futr team.
You can also follow the Futr blog to get insights on the landscape of chatbots and their potential use cases.