Introduction to ChatBot
Chatbots are computer programs to automate some tasks through chat messenger applications like Slack, Skype, Facebook Messenger, Telegram etc. It takes input from user and delivers output via chat interface based on conversational rules and sometimes with artificial intelligence.
ChatBots are new way of communication many businesses are adapting to interact with their customers and improve their customer experience leading to business growth. According to survey by Oracle, 80% of the businesses want ChatBots by 2020.
Messaging apps active user count have already surpassed social media user count in 2015. Now you can imagine why it is so important for businesses to include messaging platforms in their marketing strategy.
Source: BI Intelligence
What You Need To Know Before You Start?
- Basic chatbots are easy to develop and more intelligent ChatBots are difficult to develop where you may need to hire professionals to do that for you.
- There is a difference between Bots, Virtual-Assistant Bots, and ChatBots.
- Bots are simply any piece of computer software used to automate tasks over Internet eg. Search engine, Malicious programs etc.
- Virtual Assistant Bots are focused on automating a specific task and do it very well. E.g. x.ai
- ChatBots are like virtual assistant but multifunctional and more intelligent e.g. Skyscanner, Allo etc. They can be broadly classified into two types as Scripted Chatbots and AI Chatbots.
- You can build ChatBot at a cost of 0 to 100K+ USD depending on the budget and need of your business.
Where To Start?
Before you start ask yourself these questions:
- Think of the objectives and decide what kind of Chatbot you want to build to achieve those objectives?
- Which tasks you want to automate?
The most crucial part of the ChatBot development is Conversation Flow (or Decision Tree). You need to think of complete structure and content.
Once you are done with the conversation flow for your ChaBot, you can start with the development part using one of the most popular chatbot development frameworks like api.ai (now Dialogflow) (by Google), wit.ai (by Facebook), luis.api (by Microsoft), Watson (by IBM). (There are many ChatBot development platforms out there but personally I liked api.ai)
To get started first you need to plan Intents and Entities for your NLP (Natural Language Processing) model.
It is very important to understand the basics. Once you are familiar with Intent and Entity you are ready start development on any of the ChatBot development framework listed above.
It is important for any ChatBot to understand context of user’s conversation and make thing easy for user. Once chatbot is developed, you should test your chatbot for key metrics shown below:
- Conversational Flow
- NLP Score
- Usability & User experience
- Bot Speed
- Bot Accuracy
- Conversation Steps
You can test your bot using this tool: http://bottesting.co/. They mail you “The performance and usability report” via mail within a day. (and it’s free!)
4. ChatBot Training
The best way to build database for training your ChatBot is using real people. Invite people to chat with your chatbot and save all inputs from every user.
You can think of the queries your customer may ask. Write them in a .txt file (each query on new line). Api.ai allows you to train your chatbot by uploading maximum 10 .txt files of size upto 3 MB each.
Chatbot Development Frameworks:
- api.ai by Google – Free (with some restrictions)
- wit.ai by Facebook – Freemium
- watson by IBM- Lite, Standard and Premium plans are available.
- luis.ai by Microsoft- Lite, 10,000 API calls free per month.
Tools to design ChatBot conversation flow
- Messenger platform design kit – Free Download
- Fully customizable Sketch UI kit for Messenger platform – Free download
Chatbot Testing Tools:
- http://bottesting.co/ – Free
If you are not sure you can do it on your own. Talk to our consultant and get started with AI ChatBot development for your business growth.