Skip to main content

Create your first Chatbot

You can use any of the supported programming languages to create your first chatbot. This documentation currently support Python, JavaScript and Golang.

note

As mentioned Before almost all of language based SDK with exception of Python are community maintained and not officially supported by Sarufi. If you find any issues with the SDK, please create an issue on the respective GitHub repository. Also If you see missing documentation or typos or bugs for any of the SDK, You're welcome to contribute

Draft Architecture​

Before we start coding, let's first draft the architecture of our chatbot. Here an oversimplified version of the architecture for our chatbot. Sarufi will handle intents training, intent inference, and dialog management. We will be handling the application logic and the user interface.

Oversimplified Architecture

Installations​

Select your language of preference to get instructions on installing SDK

Its good to use virtual environment. You can read more on Virtual Environment. So, lets create a virtual environment ➡️ activate it ➡️ install sarufi.
python -m venv chatbot
source chatbot/bin/activate
pip install --upgrade sarufi

Sign up to Sarufi.io​

To use the Sarufi.io SDK, you'll need to create an account on the Sarufi.io website. Here's how:

  1. Go to the Sarufi.io website and click on the "Sign Up" button in the top right corner of the screen.

    Sarufi Dashboard

  2. Choose your preferred sign-up method (email, GitHub, or Google).

  3. Follow the prompts to complete the sign-up process.

  4. Once you've signed up, log in to your Sarufi.io account.

Obtaining Your API KEY​

After logging in to your Sarufi.io account, you can obtain your client ID and client secret by following these steps:

  1. Go to your Sarufi.io profile settings page.

  2. Scroll down to the Authorization section.

  3. Here, you'll find your API KEY.

    Sarufi Authorization

  4. Copy these values and use them in your Sarufi.io SDK client configuration.

That's it! With your API KEY, you can now use the Sarufi.io SDK to interact with the platform.

from sarufi import Sarufi
sarufi = Sarufi(api_key='your API KEY')

Create a Chatbot​

Now that we have successfully authenticated with Sarufi, we can create an empty chatbot, with 0 intents and 0 utterances.

The below image shows overview of the steps it takes to create your first chatbot with Sarufi

Create Chatbot

Creating an empty chatbot​

This is the first step in creating your first chatbot. You can use the following code to create an empty chatbot.

Creating chatbot via dashboad provides a way to create either from Scratch or Knowledge Base. You will be required to fill in important details about your chatbot.

Creating chatbot using Dashboard

Lets start by creating a flow-based chatbot. When you select a language, your chatbot will be created with some default intents and flow. You can read more on how to create Knowledge base chatbot here

Start a conversation with your Chatbot​

Now that we have successfully created an empty chatbot, we can start a conversation with our chatbot and see how it responds. You can use the following code to start a conversation with our chatbot.

Here is an example of how to start a conversation with our chatbot.

✅ To start conversation with a bot you can use one of the methods below
Class nameMethodArguments
Sarufi chat() bot_id , chat_id and message
Bot respond() chat_id and message

By default, chat_id is set to a random UUID. You can pass your own chat_id if you want to continue the conversation with the same user.

Here are examples on each method

  >>> from sarufi import Sarufi
  >>> sarufi = Sarufi(api_key='your API KEY')
  >>> sarufi.chat(bot_id=5, message='Hello')
note

As you can see, our chatbot is not able to understand the message we sent but replied us with a predefined response. This is because we haven't trained our chatbot yet. We will be training our chatbot in the next section.

Another cool way to test your chatbot is to use the Sarufi Playground. By Default when you create a chatbot, it is set to public. All you need to instantly test your chatbot is to go Sarufi Community playground and search for your chatbot.

Sarufi Playground

Adding Intents and flow​

We are going to create a bot that will help us with ordering a 🍕 pizza

Add Intent​

Let's now create an intent that will help us order a pizza and also add some utterances to that intent. You can use the following code to create an intent and add some utterances to that intent.

To make more interactive, let's general intents like greets, order_pizza, goodbye and add some utterances to those intents.

>>> from sarufi import Sarufi
>>> sarufi = Sarufi(api_key='your API KEY')
>>> chatbot = sarufi.get_bot(13)
#We assign intents attribute a dict with intent name and and intents
# dict {<intent name>:[list of all related keywords to trigger the intent]}
>>> chatbot.intents = {
...     'greets': ['hey', 'hello', 'hi', 'howdy', 'hola', 'greetings', 'good morning', 'good afternoon', 'good evening'],
...     'goodbye': ['bye', 'goodbye', 'see you later', 'see you soon', 'see you', 'talk to you later', 'talk to you soon', 'talk to you'],
...     'order_pizza': ['I need a pizza', 'I want a pizza', 'order a pizza', 'I want to order a pizza']}
2022-09-12 15:55:06,696 - root - INFO - Updating bot
2022-09-12 15:55:07,972 - root - INFO - Bot(id=13, name=My First Chatbot)
#We get bot's intents
>>> chatbot.intents
{'greets': ['hey', 'hello', 'hi', 'howdy', 'hola', 'greetings', 'good morning', 'good afternoon', 'good evening'], 'goodbye': ['bye', 'goodbye', 'see you later', 'see you soon', 'see you', 'talk to you later', 'talk to you soon', 'talk to you'], 'order_pizza': ['I need a pizza', 'I want a pizza', 'order a pizza', 'I want to order a pizza']}

Try a conversation after adding intents​

Now that we have successfully added some intents and utterances to our chatbot, let's start a conversation with our chatbot and see how it responds. You can use the following code to start a conversation with our chatbot.

>>> chatbot.respond("Hey")
2022-09-12 15:56:16,078 - root - INFO - Sending message to bot and returning response
2022-09-12 15:56:17,279 - root - INFO - Status code: 200
2022-09-12 15:56:17,279 - root - INFO - Message sent successfully
{'message': 'Intent for this message is missing in [FLOW]'}
note

You might wondering like but we trained it, yes ofcourse we trained it but we haven't given a flow on how to respond to the user. We will be adding a flow in the next section.

Lets add a flow​

Flow is a directed graph that defines the conversation flow between the user and the chatbot. Flow is defined using the JSON format. where each node is an intent and each edge is a response to the user message & the next intent to be triggered.

note

You can have nodes that are not intents. In that case, the node will be treated as a response to the user message. If you want to end the conversation, you can add a node with the name end.

Let's now add a flow to our chatbot. You can use the following code to add a flow to our chatbot.

>>> chatbot.flow = {
...     'greets': {
...         'message': ['Hi, How can I help you?'],
...         'next_state': 'end'
...     },
...     'order_pizza': {
...         'message': ['Sure, How many pizzas would you like to order?'],
...         'next_state': 'number_of_pizzas'
...     },
...     'number_of_pizzas': {
...         'message': ['Sure, What would you like to have on your pizza?'],
...         'next_state': 'pizza_toppings'
...     },
...     'pizza_toppings': {
...         'message': ['Cool, Whats your address ?'],
...         'next_state': 'address'
...     },
...     'address': {
...         'message': ['Sure, What is your phone number ?'],
...         'next_state': 'phone_number'
...     },
...     'phone_number': {
...         'message': ['Your order has been placed.', 'Thank you for ordering with us.'],
...         'next_state': 'end'
...     },
...     'goodbye': {
...         'message': ['Bye', 'See you soon'],
...         'next_state': 'end'
...     }
... }

Conversation with our chatbot after adding flow​

Now that we have successfully added a flow to our chatbot, let's start a conversation with our chatbot and see how it responds. You can use the following code to start a conversation with our chatbot.

>>> chatbot.respond('Hey')
{'message': ['Hi, How can I help you?']}
>>> chatbot.respond('Bye')
{'message': ['Bye', 'See you soon']}
>>> chatbot.respond('I want to order pizza')
{'message': ['Sure, How many pizzas would you like to order?']}
>>> chatbot.respond('3')
{'message': ['Sure, What would you like to have on your pizza?']}
>>> chatbot.respond('tomato')
{'message': ['Cool, Whats your address ?']}
>>> chatbot.respond('123, Main Street, New York')
{'message': ['Sure, What is your phone number ?']}
>>> chatbot.respond('+255783483934')
{'message': ['Your order has been placed.', 'Thank you for ordering with us.']}
note

You can see that our chatbot is able to understand the user message and respond to the user message. This is because we have trained our chatbot with intents and utterances and also added a flow to our chatbot.

Conclusion​

In this tutorial, we have learned how to create a chatbot using Sarufi. We have also learned how to train our chatbot with intents and utterances and also how to add a flow to our chatbot. We have also learned how to start a conversation with our chatbot and how to respond to the user message.