Documentation

Documentation

  • Conversation
  • Reader
  • Speech
  • Console
  • AI Agents
  • Languages iconEnglish
    • Tiếng Việt
    • Janpanese

›II. Building Chatbot on FPT.AI Platform

Documentation

  • Overview
  • I. Chatbot Introduction

    • What is Chatbot
    • Chatbot's purposes
    • Popular types of Chatbot
    • Benefits of using FPT.AI Chatbot
    • Starting with FPT.AI

    II. Building Chatbot on FPT.AI Platform

    • Working Mechanism
    • Bot building process
    • Creating your Chatbot
    • Natural Language Processing - NLP
    • Answer scenarios
    • New Bot scenarios - Bot Builder
    • Answer cards
    • Step connecting buttons
    • Getting customer's information using Form cards
    • Variable settings and management
    • Variable condition review
    • Memories
    • Send/get information via JSON API card
    • Reminder
    • Advanced functions
    • Persistent menu

    III. History

    • History - Updating Bot knowledge

    IV. Multi-channel Integration

    • Integrate with Facebook
    • Integrate with Facebook at Work
    • Integrate with Viber
    • Integrate with Zalo
    • Integrate with website
    • Webhook

    V. Broadcast

    • Broadcast

    VI. Auto Comment

    • Auto comment on Facebook

    VII. Automation Test

    • Check base on testcases
    • Auto create testcases

    VIII. Analytics

    • Introduction
    • Number of user's requests
    • Top matched intents
    • Chat with Bot
    • Number of new users
    • Top scenarios
    • Number of messages bot answered
    • Number of user's requests per hour
    • Number of active users per hour
    • Real-time analytics
    • Broadcast report

    X. Rating and Survey

    • Rating
    • Survey

    XI. Bot Setting - Management

    • Introduction
    • Bot information
    • Messages management
    • Data management
    • Bot intent confidence
    • Delete Bot data
    • Delete bot
    • Transfer bot ownership
    • Version

    XII. Bot Roles

    • Bot roles

    XIII. Chatbot SDK

    • Chat Bot SDK
  • Appendices

Live Support

  • Introduction
  • Customer list interface
  • Customer support flow
  • Note feature
  • Suppoter chat
  • KPI report feature
  • Ticket config
  • Ticket report
  • Archive conversation
  • Ticket management
  • Auto assign to supporters

Tutorials (Videos)

  • 1. Brief overview of FPT.AI
  • 2. Using QnA feature to create a Chatbot
  • 3. Creating Chatbot with complicated scenarios
  • 4. Precondition fucntion in Scenarios
  • 5. Random answer function in Scenarios
  • 6. NLP feature
  • 7. Exact match and Predictive Match in NLP
  • 8. History of Recognition
  • 9. Broadcast feature
  • 10. Live Support feature
  • 11. Persistent Menu
  • 12. Auto-comment feature on Facebook
  • 13. Bot roles
  • 14. Bot configuration in Settings section
  • 15. Integration Chatbot with Facebook
  • 16. Integration Chatbot with Facebook at Work

API Reference

  • Introduction
  • General
  • Natural Language Processing

    • Intent
    • Entity
    • Sample
    • Train
    • Predict
    • Keyword
    • Dictionary

    Dialogue Management

    • Get Answer
    • JSON Card
    • Types of the message

Building Chatbot on FPT.AI platform

Answer scenarios

After training the Bot with knowledge in your desire fields, you will need to design answers and create conversational flows for the Bot.

Scenario is a topic where the bot base in answers one. Step is a sub-topic. A scenario may include multiple steps in the same topic. These steps may be connected by buttons.

FPT.AI Conversation’s answer scenarios design system already include Default scenarios with steps “Welcome” and “Default Answer”.

  • Welcome is the greeting step, where the bot begins all conversations.

  • Default answer is the step that occurs when the Bot does not understand customer utterances. Here, the bot with respond with a default answer that is set up in this step.

You will need to edit the sample answers in these 2 steps.

You will need to design answer scenarios for different intents. For intents without entities in samples, Bot will only need to give 1 answer. However, for intents with object entities in samples, the Bot will need to give different answers based on those entities. In these cases, we will utilize variable conditions.

There are 2 methods to creating a completed step:

Method 1: Manual creation

Step 1: Create Scenarios and Steps:

Go to Scenario > Click Add scenario > Enter Scenario name > Click Create

1

2

Step 2: In bot answer condition, select Intent with matching Samples generated in NLP, or create new intents and add samples at Input.

Step 3: Drag and drop cards to design the answer. Description of these cards can be found here.

3

Step 4: You can add BUTTONS to redirect to the other Steps (Details found here).

Step 5: If you added new Intents and Samples, then Train the bot before getting to review. If you already created data in NLP beforehand, then skip this step.

Step 6. You can review if the bot answer in accordance with the designed conversational flow by clicking on the Livechat symbol to chat with the bot.

4

Method 2: Import from QNA

You can quickly import a list of Intents, Samples, and Answers into the Bot with this method, which is more suitable for answers in text forms.

Step 1: Click Import from QnA.

Step 2: Click Download sample file, then enter Intents, Samples, and Answers in the format of the downloaded file.

5

6

Step 3: Upload the file with updated information by clicking on Browse > Enter

7

Step 4: Review the results. The system had successfully entered the list of Intents, Samples, and Answers into the bot. Duplicated and invalid Samples will not be entered.

Step 5: Select Train to teach the bot the new data. Then, you can review the bot answers by clicking on the Livechat symbol to chat with the bot.

Copy step

In cases where you want to copy a created Step to another, follow these steps:

Step 1: Choose the Step to be copied -> Select Copy step to copy the step’s data.

8

FPT.AI Conversation

Step 2: Create new step in the Scenario that need the extra Step.

9

Step 3: Enter the New step name

Step 4: In Step data, use short keys “Ctrl + V” to paste the copied step.

10

Step 5: Select Create to finish.

Notice: Copy step only allows copying of the answer in the step. After copying, you will have a new step with its answer in the same format, and only need to edit its content to match with Samples. This will make the setup process faster and easier.

← Natural Language Processing - NLPNew Bot scenarios - Bot Builder →
  • Method 1: Manual creation
  • Method 2: Import from QNA
  • Copy step
Conversation
DocumentationAPI ReferenceTutorials (Video)
Reader
DocumentationAPI ReferenceTutorials
Speech
DocumentationAPI ReferenceTutorials
Copyright © 2025 FPT Corporation