What are the key differences between chatbots and virtual agents?
On paper, bots, chatbots and intelligent virtual agents might seem quite comparable… until you start communicating with them. Because mastering the human art of conversation requires a very particular set of skills.
To spot the differences between a chatbot and an Intelligent Virtual Agent, here are 4 key skills to look for.
Understanding human language
A quick reminder of “Artificial-intelligence history” might be needed here. Among the most remarkable early chatbots are two robots called ELIZA (MIT, 1966) and PARRY (Stanford, 1972), built to simulate typed conversation.
ELIZA’s program had been written by a psychiatrist to simulate a therapist, while PARRY simulated a person with paranoid schizophrenia. ELIZA and PARRY “met” several times, at the initiative of the scientific community, resulting in enjoyably absurd conversations.
Those chatbots were not capable of really understanding the human language. Because ELIZA and PARRY merely create an illusion of language understanding, by looking for keywords in the conversation to reply with preprogrammed answers.
Intelligent Virtual Agents are not based on the same type of programming as those “basic” chatbots, still employed today for instance to assist e-commerce customers.
Virtual Agents use natural language processing (NLU/NLG), pattern recognition, and contextual analysis to make more intuitive leaps, perceptions, and judgments for better speech recognition. A Virtual Agent’s conversational AI skills enables it to understand the intent of the end-user, ask additional questions to understand a context and orchestrate multiple actions.
Mastering written and vocal conversations
Even if human interactions have evolved due to the digital transformation to include more written exchanges, with emails, social media platforms and now digital workplace environments, vocal conversations with a human agent are still a deeply rooted natural human reflex.
When faced with a problem, humans still prefer having a telephone conversation rather than search for a solution online. In a service desk environment for instance, 72% of users would rather call their IT help desk than use self-serve solutions like a portal or a live-chat.
That’s why AI powered voice technologies (voicebot/callbot) are major evolutionary steps between basic chatbots and Intelligent Virtual Agents. Understanding and conversing with a human being, in different languages, and replying with a human sounding voice requires not only natural language processing (NLU/NLG) technology but also the integration of Text-to-Speech and Speech-To-Text with Multilingual voice scenarios.
For users to interact with Virtual Agents through voice in the most natural way, Konverso partners with leading VoiceAi solution providers.
Learning new skills
The evolution of an AI solution is key for the return on investment of a project.
Even if a basic chatbot can answer many of the needs identified at the start of a project, it will lack AI, deep learning and back office functionality to stay relevant and even improve over time.
Virtual Agents are more advanced artificial-intelligence solutions, capable of integrating more knowledge through existing Knowledge bases (for instance from Upland RightAnswers, ServiceNow, Confluence, etc.), and from new data sources inside and outside the company.
For instance Knowledge from existing FAQ but also unstructured knowledge from web pages, pdf documents, transcripts…
Learning new skills can also mean integrating and connecting with other technologies. For instance, a Virtual Agent can integrate with automation technologies like Robotic Process Automation (RPA) to bring more context, knowledge and personalization to automation.
If you are interested to know more about this notion of “cognitive process automation”, that really sets apart chatbots and Virtual Agents, here is a more detailed article.
Virtual Agents are also capable of automated learning based on human conversations. The more people interact with a Virtual Agent, the more they train it. As a result, Virtual Agents will constantly adapt to the user experience and the evolution of the workplace tools and methods to help maximize productivity.
Partnering with humans
Those learning skills also capitalize on human experts who will monitor the behaviour of the virtual agent and the feedback of the employees interacting with them. This partnership between an Intelligent Virtual Agent and a human requires a powerful back office, to access the interactions data but also program new abilities.
That’s why the most advanced Virtual Agents, like Konverso’s, integrate a no-code / low-code Backoffice, are easy-to-use for both IT experts (notably the Service Desk experts) and Field experts .
As a result of this seamless cooperation between IVAs and humans, Intelligent Virtual Agents can really be considered like a “digital workforce”, working alongside the human workforce.
This list of skills is based on Konverso’s experience since 2017, building a conversational AI platform and Virtual Agent solutions. Konverso has been named a top player in Everest Group's report on Intelligent Virtual Agents, both in 2019 and 2020, a recognition of our continuously improving capabilities in the field of natural language processing and Automated Learning.
About Konverso’s back Office
Konverso provides an “off the shelf” Intelligent Virtual Agent that can generate instant value for our customer. But our Back Office is also crucial for our clients to be able to extend the capabilities of the Virtual Agent.
With Konverso’s GUI based back office, our clients can personalize their Virtual Agent’s user interface. In our “drag and drop” workflow environment, they can also easily design, build, test and integrate their own conversation flows, intents and related actions.
Konverso also integrated powerful analytics on the Virtual Agents behaviour and its interactions with the users. Domain experts can train the Virtual Agent based on end users feedback, likes and dislikes. Experts can also use customers data to feed the Virtual Agent’s Machine Learning capabilities and further improve its classification performance.
To illustrate the performance of Konverso’s Intelligent Virtual Agent, here is what we achieved with Computacenter, a global provider of IT infrastructure services with 3 000 service desk agents in the world.
When Computacenter deployed Konverso’s conversational AI platform, they opened their knowledge base to let the Virtual Agent learn by itself. They also gave it access to one year of users’ interactions with the service desk. As a result, our natural language understanding rate reached 80%, even before it became available to end users. (Full Computacenter case study)
With Konverso’s intelligent virtual agents, our clients increase end-users satisfaction by 80%.