When you look back through the demo chatbots we’ve showcased in our blogs and podcasts you’ll spot a pattern – they are all led by questions and buttons.
The most straightforward way of building a bot is to pose a question to the user – such as “How can I help you?” and give the user two or three choices, such as “See menu” and “Find location”.
It’s a very tight system where the user is funnelled down a conversational route and has to click buttons to find what they’re looking for. It’s quite similar to being on the phone ringing a large company and having to press numbers to get to the right department.
What’s the problem? It’s not very ‘human’ and can be quite a slow process, especially if the user never gets the answer or information they’re looking for.
What’s the alternative? We can use a chatbot system that’s entirely conversational and lets the user ask questions and wait for a response. Not from a human, but from a bot!
The rise of the smart bots
If you’ve ever seen films like Terminator you’ll be wary of robots and AI taking over the world. With machine learning, they’re certainly one step closer to this being reality. Google has very much pioneered the way with machine learning. They have spent a lot of time creating a system that recognises human interactions and can mimic and replicate human experiences. This is the sort of technology we can use with a smart bot. By connecting our bot with Google’s Dialogflow we can harness the power of their machine learning and build a truly authentic AI experience. With a smart bot, the user doesn’t need to press buttons, but can type a question, such as “When are you open?” Within Dialogflow you can give it a range of questions the user might ask and it can understand what they are asking and offer an answer. Using Dialogflow the bot will reply to the user’s question by saying “We’re open between 9am and 5 pm every day.” What happens, though if the user says “Are you open on Monday?” Would that be too different and too confusing? With training, your bot can understand and interpret basically any question the user might have that relates to your business.
Staff training for your bot!
When a business takes on a new member of staff you need to invest some time in training the new member of staff to undertake their job successfully. A chatbot is very much a new member of staff. It too needs to be trained. Going back to our previous question – “Are you open on Monday?” The bot can reply again with, “We’re open between 9am and 5pm every day.” Once you offer the bot the question about Monday and train it to reply with the sentence about being open every day, it’s sorted. What happens, though, when you ask it about Tuesday or Saturday? Within Dialogflow it uses machine learning to understand that Monday is a day of the week. So it knows if you ask if you are open on Saturday this is the same as asking about Monday. Therefore, it will offer the same answer as it did for the Monday question. You don’t need to plan for every single question a user might have, but offer a range of questions and let the AI learn. Sound complicated? It’s really not. All you have to do is offer a range of questions the user might ask and the most appropriate answer to give to these questions and the machine learning does the rest. Much like training a new receptionist to use the email system at work correctly – they will use their experience from previous jobs to make it easy to understand and learn. All you have to do is take some time to train your bot to do its job correctly.
Live chat, 24/7
I’ve mentioned in previous blogs about how chatbots are very much the future of modern business. Every business will have one and need one – much like how they currently need a website and social media. What makes a chatbot better than what they currently have? A chatbot is live 24/7, 365 days a year. Going back to our analogy about training a new receptionist. They will only work set hours each day and will go home at some point and customers can’t ring or email and get help or answers. Unless you’re running a business that generates millions each year, you won’t be able to afford the staff to work 24/7. What happens if the receptionist takes a holiday, unfortunately, becomes unwell or takes maternity/paternity leave? Bots don’t take holidays, get ill or get pregnant (not yet, anyway). So they will be grinding away every single day. With a smart bot, it can quickly and easily answer customer’s questions on a variety of topics and boost customer confidence in the business.
Combining the chatbot types
I’ve mentioned at the start that the ‘questions and buttons’ model of a chatbot is limited. Am I advocating moving towards an entirely AI-led chatbot?
What we offer at Tiny Blue Rocket is a ‘best of both worlds’ bot. Our chatbots are built primarily to offer questions and buttons to the user to help them see what the bot can do and help them find what they’re looking for in a method similar to a website.
Additionally, we integrate Dialogflow and it’s machine learning into our chatbots. So if a customer can’t find what they’re looking for from the questions and buttons they can type away and get a helpful response.
Essentially, we build layered bots that serve all sorts of user interactions, both simple or complex.
By the end of 2020, around 80% of all businesses will have a chatbot. When you decide to get a chatbot for your business make sure it’s a smart bot!