Chatbots wait, they wait for a visitor to open the chat interface and start a conversation. Day in and and day out, the chatbot's life has a large portion of waiting and watching while visitors click other buttons, fill up forms, select text or navigate through various pages looking for information. Well, the chatbot has the information and can certainly respond better than Mr.Searchbox in the header! But, how many people would want to have a conversation with a chatbot? They know it will be robotic and they know that it is a fancier Mr. Searchbox in many ways!
With v2.0 of Illumea, we introduced a new feature called "Pro-active Engagement". While the implementation of this is mostly smart thinking and good JavaScript usage, the idea is to give the chatbot the ability to make a first contact with the visitor while keeping context. If a visitor is on a pricing page, the chatbot might proactively surface a callout saying - "Hey! do you want to compare the pricing for these 3 products you are looking at?" or "You get 20 features at a USD 100" and if they move to a product page, the chatbot could ping something like "Looking for compatibility details? I can help you check". Here the chatbot is no longer just responding to intent, it is helping create it. These proactive callouts grab attention and maybe start a conversation that could lead to a sale even. For returning visitors, the conversation memory of Illumea enables the chatbot to tie in past conversations and converse with a more natural flow as a person would.
Today, we are at v2.6 of Illumea and we are continuing to improve the way the chatbot interacts, adding improved and tunable behaviour, personality presets, support for emoticons and quick in-context jokes that make the conversation feel more natural and less robot-like. Personality and behaviour are two key areas we are continuing to work upon.
Imagine walking into a wildlife resort and speaking to a receptionist. You would expect a different conversation than you would have with a technical support engineer of a software company. The information might come from the underlying knowledge base, but the way it is delivered should be different. The chatbot can be configured to behave differently based on the role and we are working on auto-detection as well, enabling the behaviour to switch as the conversation progresses. While interaction is an easier part, enabling the bot to make decisions based on the behaviour is the tricky bit. You could easily get a chatbot to answer a question, but, making it detect something like a buying signal changes the game!
Most chatbot development is focused on knowledge to ensure that it can answer questions correctly, but, knowledge alone does not make an experience feel intelligent. We want to make the website a two-way, intelligent experience making the conversational layer more adaptive.
We want to borrow the things that make human interactions effective like awareness, context, timing, character, judgement and purpose. Come to think of it, we are human who are building a chatbot to behave more like a human to help other humans.