What is a chatbot?
A chatbot is a piece of software that has been programmed to recognize and respond to human speech — mimicking a conversation between two people. Chatbots are rules-based, meaning they are designed to understand and respond to selected keywords or phrases. When a person uses a keyword that is recognized, the chatbot replies with a preset answer.
In customer service, chatbots can be used to answer FAQs or carry out simple tasks like placing orders or offering recommendations. Publishing company Harper Collins, for example, has a chatbot that finds customers their next read, based on users’ answers to questions about their tastes and preferences. And fitness brand Peloton uses an entirely button-based chatbot to help solve customer queries around membership, placing an order, returning items, and other issues.
What is a virtual agent?
A virtual agent (also known as an intelligent virtual assistant, or IVA) is a software program that uses artificial intelligence to recognize human speech in the way it’s really used. As with chatbots, virtual agents simulate human conversations. With the help of AI, these bots are able to engage with people in a natural way.
IVAs are also used by customer support teams to automate repetitive tasks — but that’s pretty much where the similarities end. In terms of complexity, virtual agents leave rules-based chatbots in the dust.
Virtual agents vs. chatbots
Chatbots and virtual agents are both designed to facilitate bot-to-human conversations. And within the field of customer support, both can be used to automate simple tasks and resolve repetitive queries — freeing up agent time to work on more rewarding cases.
So what’s the difference?
Virtual agents are powered by conversational AI
People don’t communicate with each other using only a limited set of words and phrases. We talk in text-speech and colloquial phrases, our writing is filled with typos and abbreviations, and there multiple different ways of expressing the same sentiment. This makes chatbots’ reliance on keywords frustrating.
Say you wanted to check if your favorite airline flies to the Moroccan city of Marrakesh. If you asked their chatbot using the (still correct, but less commonly used by English speakers) spelling “Marrakech” it might not understand your question.
To avoid bot confusion — and human frustration — many rules-based chatbots guide people through a dialogue flow using buttons. This decision-tree model gives users a small number of answers to choose from. With their limited ability to understand natural human language, chatbots are best suited to taking on simple tasks where a small amount of information is required.
IVAs, on the other hand, are enhanced with conversational AI: including natural language processing (NLP) and natural language understanding (NLU). This allows virtual agents to establish what a user is really asking (or their intent) regardless of the exact wording they use.
Because these AI chatbots are programmed to understand the overall meaning of a text — rather than specific words — they won’t get tripped up by synonyms, typos, or abbreviations. You can ask a virtual agent open-ended questions, and from the context of the conversation they’ll be able to give you a relevant answer.
More on conversational AI
IVAs are enhanced with machine learning
Powered by AI, virtual agents have the ability to learn and improve. As IVAs take in new data, they use machine learning to get better at recognizing the various ways different intents are expressed. This means these clever bots become more accurate over time.
Chatbots can only recognize words or phrases they have been specifically programmed to understand. So no matter how many times you ask a chatbot if there are flights available to “Marrakech” — instead of “Marrakesh” — the bot won’t learn that you’re talking about the same city.
If the chatbot can’t understand what a customer is saying, not only is this a frustrating experience, but a human agent still has to get involved to resolve the issue.
Chatbots are simpler to set up
One of the benefits of using a chatbot, is that while they are less sophisticated than virtual agents, this means they’re relatively easy to set up, and cheap to maintain. Due to their complexity, it takes a little longer (and requires more resources) to get a virtual agent running smoothly.
Are all virtual agents the same?
Unfortunately it’s not as simple as choosing between a chatbot and an IVA… Because not every virtual agent platform has the same capabilities.
Here are some of the features you should look out for when deciding on an automation solution for your customer service (spoiler alert: Ultimate’s powerful virtual agent platform has all these features, and more).
1. Omnichannel automation. According to research by Forrester, email is currently the number 1 support channel for customer service. But customers expect to be able to reach out to brands however they please. The best automation providers will offer omnichannel solutions, automating customer support requests across any digital channel: covering email, chat, and social media or messaging apps.
2. CRM integrations. For customer-obsessed CS teams, a headless automation platform (meaning one that sits inside your existing CRM — just as a human agent would) makes the most sense. Not only does this save your team from having to learn how to use an entirely new system, but it allows your virtual agent to access customer data and personalize interactions. As 80% of consumers are more likely to buy from brands that offer personalized experiences, the headless approach is a no-brainer.
3. Multilingual AI. If you want to scale your customer service around the world, a multilingual virtual agent is a must. Multilingual AI allows companies to expand into new markets and serve different timezones, while offering localized support at native-level fluency — so your bot will grow alongside you.
4. A no-code platform. No-code platforms are designed to be intuitive, making them simple to use and maintain. Since no-code solutions are accessible to non-technical users, you won’t need to invest in additional IT support, and it’s easy to onboard new bot managers. Using a low-code platform, on the other hand, requires an understanding of programming languages. This means low-code solutions take longer to set up, and you’ll have to hire a developer to take care of the automations.
5. Custom solutions. While out-of-the-box automations are faster to implement, creating a custom-built solution using API integrations will allow you to fully automate more processes.
At Ultimate, we understand that every customer service team has different needs, that require personalized solutions:
“The team at Ultimate had an immediate grasp of our issue. They provided us with a made-to-measure, integrated solution to automatically tag our tickets and implement automated replies.”
- Ilaria Vilardi, Customer Service Senior Specialist, Clue
The type of automation solution you choose will depend on the particular needs of your CS team. But regardless of whether you go with a chatbot or virtual agent, automating your support will take care of simple queries, allow you to serve customers around the clock, and free agent time to work on high-value interactions.