What is conversational AI? Our in-depth guide

Looking for a definition of conversational AI – as well as examples of how it's used by major brands today? Look no further.

February 19, 2021
What is conversational AI? Our in-depth guide

Every time you tell Alexa to put on your favorite song, you’re engaging with conversational AI. Ever asked Siri for directions? That’s also conversational AI at work. Talked to a chatbot online recently? You’ve guessed it: conversational AI.

So, what is it, exactly? Conversational AI refers to any technology that can mimic human conversational interactions, drawing upon machine learning and natural language processing (more on these later) to recognize your speech and text. Once it has interpreted what you’ve said and what you mean, it has the ability to respond in kind.

Any search engine results page can give you a textbook definition for conversational AI, but what they might not tell you is how much the industry is growing. The conversational AI platform market is expected to be worth more than $17 billion by 2025, growing by roughly 30% each year until then. We'd like you to remember this stat for now.

That’s why now is a good time to get ahead of the game and learn the ins and outs of conversational AI.  

No, really, what is conversational AI?

Okay, so we’ve covered some of the basics – conversational AI is software, platforms or other tools that you can talk with in two-way dialogue. Through the use of natural language as the interface, users can find information, make a transaction or trigger an event, like playing music in a smart home device.

But let’s root around under the hood and explain more about how they do that, starting with the ‘AI’ bit.

Conversational AI and machine learning

The history of artificial intelligence (AI) encompasses all efforts to recreate human intelligence in machines. So it’s a broad subject to define. If you can program a computer to solve problems, perform actions and make decisions based on its environment and external inputs, you’re dabbling in AI.

However, conversational AI goes a step further. Having a realistic, two-way conversation with a human requires more than pre-programmed rules and responses. This is where machine learning comes in.

Machine learning is a subset of AI that specifically refers to technologies that can learn by themselves (in a strictly supervised way). It is programmed with the rules and pattern-finding requirements to make informed decisions, without those specific decisions being programmed and solutioned for individually. They learn from their mistakes, too, which is crucial when dealing with the weird and wonderful idiosyncrasies of human language and speech.

More sophisticated conversational AIs may include elements of machine learning, although it does not necessarily have to.

You may notice small changes in the way Siri or Alexa answer questions, for example, as they use machine learning to constantly adapt to find what it determines to be the right answer.

Other conversational AI err on the side of caution. Some healthcare chatbots, meanwhile, may not use machine learning, instead opting to use prescribed answers to potentially life-or-death user requests.

Natural language processing

Natural language processing – or NLP – methods can recognize inputs, analyze language and then provide an appropriate output.

Put simply, NLP is how a computer listens, interprets and then responds as a human would. It’s somewhat like the brain of a conversational AI.

What’s required for NLP to work is an input. Users can type to a chatbot, speak to a voice assistant, and some conversational AI have even been trained to recognize sign language.

The NLP then translates this input data and tries to find intent. From there, it will choose the best response for the conversational AI interface (be it a chatbot, voice assistant or digital human) to give.

Types of conversational AI

We gave a quick summary at the beginning of this article, but here are a few examples of the types of conversational AI you might interact with:

  • Voice assistants: Technologies that can understand your spoken commands and also respond verbally are often called voice assistants or virtual assistants… Or intelligent personal assistants… They go by a lot of names actually; but we’re essentially talking about the Siris, Alexas and Google Assistants of the world.
  • Chatbots: If you don’t want to interact verbally, chatbots offer a text-based alternative to voice assistants. Type in your questions or commands and they’ll respond with the information you need. They’re quick and simple, but chatbots often lack the personality that comes with a voice assistant.
  • Digital humans: A digital human is the next level in conversational AI technology. They take the best features of chatbots (text) and voice assistants (speech), combine the two and add a face and personality to the mix. Studies have shown that AI with a face can help develop rapport and support more empathy-driven interactions between humans and machines.
  • Large language models (LLM): In truth, LLMs power the above technologies without being a seperate class. But they've exploded onto the scene so heavily, they're worth mentioning as a form of conversational AI. These generative AI models create humanlike dialogue in real time thanks to their ludicrously massive data set.

So “why” conversational AI?

Consumers miss the human touch. A PwC study revealed that 59% of people believe companies have lost the human element of their customer service. A huge 82% say they’d now rather talk to a human than with automated, robotic technologies.

This puts brands in a tricky spot. Their staff can’t be everywhere at once, so some automation is necessary to provide important services. But it can be difficult to form lasting connections with customers and deliver memorable experiences through technology alone.

Done well, conversational AI is the solution to this “digital customer fatigue”, providing the advantages of digital through a natural interface. Brands can enjoy scalability, cost efficiencies, better engagement and a smoother sales process, while customers can receive answers to their questions naturally, with low effort, quickly and efficiently, 24 hours a day.

People also come away with a feeling that when they talk, your brand will listen and respond. No lifeless automated email responses or endless hold music.

Indeed, our research has found that almost half (47%) of brands consider customer satisfaction as their most important metric for measuring the success of a chatbot strategy, rather than cost or efficiency.

These add to some other important goals, including reducing operating costs, improving the number of customer interactions the brand can manage and resolving customer issues quickly.

Using conversational AI in business

One of the reasons the conversational AI market is growing so rapidly is that development costs for chatbots are dropping and more businesses are beginning to recognize there are strong omnichannel deployment opportunities.  

Digital workforces incorporating chatbots, virtual assistants and digital humans are being used across many industries, including:

  • Healthcare
  • Financial services
  • Retail
  • Education
  • Technology and software
  • Property management
  • Entertainment
  • Government and public services

Many of the commercial applications of conversational AI are overlapping between industries. So instead of breaking down each industry, let’s look at some of the popular conversational AI use cases being deployed today.

Customer support

Automating customer support functions is probably one of the first use cases that spring to mind when you think of conversational AI platforms. Gartner predicted 85% of all its customer interactions with a brand would be through these technologies by the end of last year. You’ll no doubt have already encountered a customer support chatbot online before while browsing the web.

Conversational AI in retail, for instance, can help steer users around a website, answer frequently asked questions, provide 24/7 support and hand customers over to a human representative when necessary.

Personal assistants

Apple, Amazon and Google are investing a lot in their voice assistant technology, and it seems to be paying off. In 2019, 53% of people were already using a voice-activated personal assistant at least once a week, according to NPR research. And this figure climbed even higher to 63% last year.  

The study also revealed that 53% of people who use voice assistants have them turned on permanently and one-fifth speak to their conversational AI several times a day.

Many of the emerging use cases for using conversational AI in business stem from this personal assistant space, too.

Sales, marketing and business development

Conversational AI is a great support tool for sales and marketing teams. In fact, we wrote a whole article on the subject last year. Whether it’s lead generation, business development or after-sales support, conversational marketing is helping brands make the most of every sales opportunity.

And it’s not just about convenience or scalability. When polled, 55% of marketers said AI-powered web experiences improve customer experience and engagement. A similar percentage confirmed they boost conversion rates.

One example of this put into practice is when conversational AI meets financial services. Digital humans working in banking or mortgage industries, for instance, are helping first-home buyers learn more and fill out disengaging loan application forms. It’s sales and business development, but it feels like conversation. Because digital humans have all the time in the world to dedicate to each potential customer, they can help nurture leads.

Concierge services

Booking hotels, filling out forms, paying bills – life is full of tedious, time-consuming or just plain confusing tasks. Many brands are now using conversational AI to provide concierge-type assistance to customers completing life’s little mundanities.

Even quite complex tasks are getting the conversational AI treatment, such as guiding people through the mortgage documentation process.

What does the future of digital human concierge services look like, we hear you enthusiastically ask. Here’s a concept we created for the City of Darwin in Australia, to show what can be done on a grand, city-wide scale.


Conversational AI has huge potential in healthcare to help revolutionize the way services are provided.

For example, a digital human is capable of having in-depth conversations with elderly patients or people with dementia, providing medical information and crucial companionship. Similarly, AI technologies are available during mental health emergencies, making sure people always have someone to talk to when they need it most.

Mentemia is a healthcare app co-founded by New Zealand’s rugby great Sir John Kirwan. We teamed up with them to build the world’s first digital human sleep coach, capable of providing companionship and a personal plan for better sleep, all delivered through Sir John’s digital twin.

What’s the future of conversational AI?

We’ve mentioned that conversational AI platforms are set to become a $17 billion market by 2025. But here's where we tell you the industry has already far outstripped that. In 2023, OpenAI – the providers of the most impressive and perhaps most powerful language model – is worth $29 billion on its own.

Few could have predicted how conversational AI could have captured people's imaginations so quickly. Building on the market penetration of Siri and Alexa, ChatGPT and similar technologies are set to change how the world moves forward with conversational AI.

But something is still missing.

We think digital humans will have a significant place in that market, because they’re the only interface capable of replicating the personalized human touch people want.

ChatGPT might feel like text chatting a real person; but it doesn't replicate the feeling of being in real conversation. Whether we (as users) are patients, staff or customers, we all wish to be seen, heard and valued.