“Hey, what’s your opening hours?”
“Sorry, I don’t understand”
“Excuse me, where are you located?”
“Can you repeat that”?
As humans, we’re used to talking to other people in a way they can understand and respond accordingly. Thus, if they couldn’t understand our purpose, the conversation would be useless. And so, when designing a conversation, the Chatbot’s Intents are one of the most important building blocks: after we direct the conversation to a certain flow, we need not only to get the user’s reaction – but to understand what he meant.
Earlier, in our chatbot building platform at CoCoHub, we started by having 4 basic intents for chatbots – Yes, No, Stop, Continue. Essentially, it means that even if the user says “Yeah” we’ll understand that he meant “Yes”, and by saying “Nope”, he meant “No”. We also enabled the possibility of using specific keywords, and having the default “Anything Else” option – in which, every user response is treated with the same follow-up.
However, we realized that those 4 intents on chatbots are never enough. Most of the time, you want to create your own intent. Whether it’s creating the same intents for a chatbot in different language, or capturing the user’s responses in a specific context – it’s a feature we’ve worked on for some time, and now, we’re proud to present our newest twist:
The Intents Revolution.
Yeah, we just expanded the list of intents to more than 50, including most popular cases in human-to-human, human-to-computer and customer-to-business interactions. For example, when user says hi or bye, asks the virtual agent for his age, gender, location or religion, tries to find out a business’ working hours, location or costs, wonders what’s the time or weather, expresses emotions, asks the chatbot to tell a joke or even, because, you know, we live in a certain reality – misbehaves, or talks inappropriately.
What’s in it for me? How can it make my chatbot better?
So. Much. Awesome. Stuff.
First of all, you no longer have to write in multiple inputs regarding those chatbot intents. We already have them there. So now, you can try it to see if it covers all major options you can think about. From there, you can customize them to your needs, add or subtract so it fits local or cultural sensibilities.
Then, all left for you to do is to create a follow-up that’s relevant to your use case or brand. So in this example, the chatbot understands that the user asks for the service’s costs, if the business has an Instagram page or where the store is located. Hence, you write only the response.
You’re the best! How do I use the new chatbots intents?
We’ve created a list of intents that you can browse – or search for specific ones, since we gave every intent one or more tags. Then you just click on it in the navigation node, and it’s there.
Now, all left for you to do is to write the follow-up response from the chatbot.
How do I find a specific chatbots intents for my use case?
That’s pretty simple!
First of all, take a navigation node, click “Add an intent” and a menu will pop up. Now you can:
- Type your desired intent in the search window, or
- Scroll down and up, up and down, until you find something fitting.
Once you hover with the mouse over the intent, you’ll see what it does and some examples of relevant inputs.
Now, by clicking on the eye button, you can get inside and see the complete variety of phrases and utterances.
I couldn’t find the chatbots intents I looked for. Can I create one by myself?
You betcha! So, just click the “Make your own” link or the “Create” button – and an intent creation menu will appear. Now:
Now you’ve got two possibilities – the learned intents, in which we put a little machine learning magic powder into the mix, and the keywords intents, for specific use cases.
Create a learned intent
- Give the intent a name. The name (with just lowercase letters and dash or underscore) will reflect the intent’s purpose or essence.
- Click “Add first phrase”, then add relevant phrases. Press “Enter” to move to a new phrase. You can always delete existing phrases. You just need to provide an initial list of keywords and phrases there, not to come up with every possible phrase that was ever uttered from the dawn of humanity. We take care of it: that’s where machine learning comes into play.
- Go to “Publishing settings” – add some relevant tags (such as “customer service”, “small talk”, “name”, etc) that will help in finding your chatbots’ intents.
- Add a description – a short explanation on the intent’s purpose (“User tries to…”)
- If the intent is specific for your purpose, business or use case, just click “save” and it’s saved. If you think it’s useful enough for all our community, click on “publish request” and we’ll review it – if it’s good enough, you’ll have your own intent in the hub!
Create a Keywords Intent
If You don’t want the machine learning to do its magic and you’re aiming for very specific keywords – just choose the keyword intent. Let’s say that you want to tell people about girl’s names, and you have a specific response for different names.
In this case, choose “create keyword intent” from the first menu of intent creating
1. Repeat step 1 from the learned keyword – give it a name, say, “name-sharon”
2. Enter specific keywords such as, well, “Sharon”
3. Go to “Publishing settings” – add some relevant tags (such as “name”, “sharon”) that will help you in finding your chatbots’ keywords.
4. Add a description – a short explanation on the intent’s purpose (“User asks about the name Sharon”)
5. Click “Save”
6. Create an answer to every specific keywords
And – you’re done! Your new chatbot’s intents are ready! Hurray!
Sounds familiar. So essentially, the intents’ system works like the components’ system?
EXACTLY. You’re good. Has anybody told you today that you’re good? If not, let me be the first. I’m really proud of you.