Voicebot and Chatbot. These terms have been cropping up more and more as AI technology finds its way into pretty much every industry.
They’re dominating the headlines, they’re gaining new capabilities, and sometimes they even pass the Turing test.
But when you scratch below the surface, you’ll find that many voicebot definitions sound just like the chatbot definition you already have open in another tab (and vice versa).
So, is the clue really in the name (i.e different names for different kinds of bot) or is one much the same as the other? In this article:
- Voicebot vs. Chatbot – how different are they?
- A Question of Conversational User Interfaces
- Do Voicebots and Chatbots have the same automation capability?
- What are Voicebots and Chatbots used for?
Voicebot vs. Chatbot – how different are they?
You may assume from the name that a voicebot is basically a chatbot that you can speak to. And you’d actually be right.
But there is more to it than that.
First things first – yes, the first difference between voicebots and chatbots is their method of communicating with the user. Both use examples of conversational user interface (conversational UI), but where a chatbot communicates with written text, a voicebot can understand and produce spoken language.
In both cases, the conversational UI allows a user to interact in a conversational way with a computer system. This is instead of a graphical interface like a taskbar or control panel.
In both cases, the human user can use this to make requests of the AI that’s connected to the UI. This will often take the form of a virtual assistant within a speaker or app.
There are also some differences in the way that these two forms of conversational AI are applied. These stem directly from their differing ways of communicating.
A Question of Conversational User Interface
Before we unpack the different Bot capabilities of voicebots and chatbots, let’s focus on what the user encounters first: voice and/or chat.
Both of these are examples of conversational UI, a form of interface that allows humans to interact with computers in a way that mimics natural conversation.
Conversational UI is considered a recent evolution in human-computer interactions, following text-based UI and graphical UI.
(Most search engines use text-based UI, whereas computer applications like Photoshop use graphical UI).
There’s another evolutionary timeline within conversational UI, where voice-enabled conversational UI is the newest iteration of the technology.
Voicebots use voice-enabled conversational UI to understand and respond to a user’s spoken instructions or questions. Chatbots do the same thing, but they use text instead of speech.
Both kinds of bot are able to do this thanks to Natural Language Understanding (NLU), which allows the AI to parse speech or writing for the user’s ‘intents’. The intent is whatever the user is trying to achieve (such as booking tickets), and once the bot understands this it can take action to help.
The difference with voicebots is the additional technology working alongside NLU to process and produce spoken language. It goes something like this:
- Speech-to-text (STT) functionality transcribes the human user’s questions, instructions or greetings into text.
- That text is analysed with Natural Language Understanding (NLU), the ‘intents’ are recognised and appropriate responses selected.
- Text-to-speech (TTS) functionality turns the chosen text responses into synthesised speech. This can be highly realistic.
It should be noted that where we say ‘responses’, we don’t just mean conversational responses. Many chatbots and voicebots can do a lot more than hold a conversation, with many boasting specialized capabilities used by businesses to provide better service.
Do Voicebots and Chatbots have the same automation capability?
The clue is in the name!
While they may communicate differently, the ‘bots’ connected to each conversational UI have access to a very similar range of abilities.
Once instructed, they can do things like:
- Retrieve information (like flight times) from other computer systems.
- Present customers with tailored messages and offers.
- Update account information on the user’s behalf.
This kind of thing will sound familiar to anyone who’s encountered the “celebrity” voicebots (Alexa, Siri, Cortana and similar) but the same technology can be used for more specialized automations within customer service.
There’s no standardization, which means that you can have different kinds of ‘Bot’ hooked up to the conversational UI. These can be classified as follows:
- Rule-based bots – these chatbots choose their response based on predefined rules in a database. They can recognize language inputs from the user, and the response will have been ‘humanly hand-coded’ and ‘presented with conversational patterns’. Many of the earliest rule-based chatbots made users choose from predefined dialogue inputs (a bit like a choose your own adventure novel).
- Retrieval-based bots – the magic word here is ‘integration’. Retrieval bots respond more flexibly by accessing other systems through APIs. This approach allows the user to query anything from college course information to flight times.
- Generative bots – this kind of AI uses deep learning to produce highly versatile and sometimes unpredictable conversational responses. Generative AI analyzes a large quantity of training data and then algorithmically produces a response that’s most likely to suit the input query. This is a bit like predictive text on your phone, but on a much larger scale.
How does this affect our understanding of the difference between voicebots and chatbots?
It means that the two different forms of conversational UI (voice and chat) can each be paired with different kinds of AI.
Chatbot A and voicebot A may have similarly intelligent AI ‘brains’. Chatbot B might look identical to chatbot A but only have access to limited resources behind the scenes.
This is perhaps a reason for the confusion surrounding the terms voicebot and chatbot; anyone who’s only encountered the most basic of rule-based chatbots might wonder what could possibly be gained by giving that technology a voice.
The answer is that because the technologies (voice recognition and integration with other systems) have moved on, voicebot deployments come with far greater capabilities than the earlier chatbots.
What are Voicebots and Chatbots used for?
Practical applications is an area where the two kinds of bot diverge, each with their own advantages.
Voicebots, often deployed to automate phone-based customer service interaction, are more likely to have the caller’s full attention, and may have to deal with irate callers. Chatbots, meanwhile, have their own set of unique tricks available.
Customer service voicebots
Their ability to handle spoken language make voicebots ideal for phone-based customer service interactions. Voicebots – unlike chatbots – can be the caller’s sole touchpoint. Where a chatbot has to share the user’s attention with the webpage it’s leaped out from, a voicebot may have to stand alone.
Callers are likely to pay more attention to the voicebot’s performance because they may not be viewing the company website at the same time. This makes accurate Natural Language Understanding highly important for voicebots.
People who access a business via this kind of UI may be in some distress (if they’re calling to make an insurance claim or a complaint.) This puts additional pressure on the voicebot to come across as professional, and to help resolve the caller’s issue promptly so that they don’t have time to notice any artifice.
To achieve this, a customer service voicebot has to noticeably improve customer service efficiency, which can be done by automation – either partial or full.
Partial voicebot automation
This is where the Voicebot answers the phone (because it is part of the phone system) and gathers caller intent data. In other words, it asks the customer why they’ve called, and then parses the caller’s reply for their ‘intent’ which it checks against a database.
These systems can also take information such as name, date of birth, policy number, or favorite color. There would be no practical advantage to that last one, but it would be possible with a voicebot IVR system that allows customization.
The voicebot then automatically hands over to a human agent, who receives all the information that the bot has collected on their screen.
Full voicebot automation
In this case the voicebot handles the entire interaction, and achieves what the caller wants without involving a human agent. Examples include taking meter readings, providing information and even updating insurance policy information (such as car mileage).
Integration with other business systems (like CRM databases) is what elevates a good voicebot above the early forms of chatbot because it allows them to complete admin tasks so the human agent doesn’t have to.
Customer service chatbots
The chatbots which live on customer facing websites can do simultaneously more and less than the average voicebot.
Deployed as a kind of ‘web assistant’, chatbots often serve to help the visitor get what they need (mortgage advice, discounted boots – the possibilities are endless) from the website.
The website’s content does some of the heavy lifting here, with the chatbot acting more as a tour guide than as a front of house agent.
This means that, a key difference in practical chatbot application (compared to the Voicebot) is reduced emphasis on realistic conversational skills, and increased emphasis on well timed behavior triggers.
Good website data is key here, because if you can narrow down where and when users tend to leave the site in frustration, you can set the chatbot to open itself up and offer to help.
Something else the chatbot can do that a voicebot can’t.
Show you things.
Occupying their half-way position between menu and assistant, a chatbot could (for example) display a map showing the position of your courier (and your discounted boots).
(Side note, does anyone remember when landlines with built-in webcams looked like they were going to be the next big thing? No, us neither.)
Upgrade your IVR with voicebot customer service
Combining partial and full automation in the contact center with a well designed VoiceBot IVR system can have a significant impact on the bottom line.
With repetitive questions taken care of, the human agents can focus on listening to the customer and solving their problems.
The customer arrives in a better mood having been routed efficiently to the human agent by the VoiceBot, and average handle time for agents reduces by 33% on average.
Click here to learn why upgrading your IVR to include voicebot technology is the best thing you’ll ever do for your customer experience.