Technical Articles
SAP Conversational Artificial Intelligence | Building Chatbot with AI | Step by Step Process of Deploying an SAP AI Chatbot on Microsoft Teams and Telegram
Introduction
SAP Conversational AI is an end-to-end chatbot-building platform that enables people to train, build, connect, and monitor intelligent chatbots fully integrated with SAP products. It helps users efficiently manage business tasks. With enterprise-ready chatbots able to guide users to the right content, answer FAQs, and automate tasks, businesses can transform the experience of customers and employees alike. In this blog, you’ll learn how to build your chatbot, create conversational flows, connect the chatbot to communication channels, and monitor its usage.
How to register at SAP Conversational AI?
Step -1. Create an account on the SAP Conversational AI Platform. (It’s free)
How to Build Chatbots with SAP Conversational AI?
This is a 4-Step process.
Step-1: We are going to create a chatbot that retrieves answers using conversational flows. This will save our time and also allow us to create custom skills for the bot.
Step- 2: Choose a unique name for the bot and write description. Also, you can select topics that will help the bot to train. The default language selected for the chatbot is ‘English’. Let’s leave it as it is.
Step -3: For data policy, I am choosing ‘Non-Personal’. You can choose any according to your requirements. As recommended for Store Conservation data, I will also choose ‘Store’.Now, choose the end-user type as ‘Non-vulnerable’.
Step -4: I will choose ‘Private’ as I don’t want it should be visible to everyone. Now, Click on ‘Create A Bot’.
Bot has been created and is ready to use. There are 4 Phases of Bot building.
- TRAIN – Make your bot understand the requirements.
- BUILD – Creating Conversation flows.
- CONNECT – Deploying the bot on various platforms.
- MONITOR – It allows the bot to improve from time to time.
TRAIN: We are required to upload a “.csv” file contains pair of questions/answers. Also, we can add language as per our requirements.
Open Notepad, On the left side of ; question has been written and on the right side, an answer, as shown below. Press enter to write another pair of question/answer.
Save this file as ‘name.csv’
Click on Upload or simply drag and drop the file to upload.
Once the file is uploaded, the file can be seen as shown below. For any changes in the file, need to enable the Edit mode.
As the Edit mode is enabled, we can see the pair of questions/answers.
BUILD: This option allows us to create some predefined skills like greetings or small talk etc.
Let’s click on Small- talk, we can see the predefined answer of ‘hello’. We can edit and set as per our requirement, even we can add the same answer of 2 same types of questions.
Once the pair of questions/answers have been uploaded, we can test the bot as shown below.
Working fine.
How to Connect AI Chatbot to MS Teams?
Step 1: Create/ login into Microsoft Azure account.
Step 2: Bot channel registration.
Step 3: Get your Application (Client) ID and secret value.
Step 4: Connect chatbot to Azure.
Step 5: Test bot in Azure Web Chat.
Step 6: Test bot in MS Team.
STEP 1: Create Microsoft Azure account
Go to Microsoft Azure and create a trial account. (If Already have a subscription, then login into the account)
After creating, make sure your subscription is set up properly.
- Go to the Microsoft Azure portal. (HOME)
- Click Subscriptions.
- Click on Subscription name. (Free Trial)
- Scroll down and click on Resource Providers (menu on the bottom left).
- Make sure Microsoft.insights is registered.
Step 2: Create Azure bot channel
Once you are logged in, Search for Bot Channels Registration in Marketplace and click on Create.
Field Name | Input Value |
Bot Name | A unique display name for the bot (which will appear in channels and directories – this can be changed later) |
Subscription | Your Azure subscription (Trial / Paid) |
Resource Group | Select a resource group. If don’t have – then create a new one. |
Location | Choose a location near where your bot is deployed |
Pricing Tier | F0 (10K Premium Messages) |
Messaging endpoint | This can be filled out later |
Application Insights | On |
Application Insights Location | Choose a location near where your bot is deployed |
Select Create. It will validate and you will get the notification of Successful Deployment. (Shown below)
Step 3: Get your app (Client) ID and secret Client Value
- Go to Home, click on Resouce Groups, Select the Bot channel you just created
- Click Configuration, and select Manage
- By clicking Manage, you will be redirected to Certificates & secrets
- Generate a Client Secret by clicking New Client Secret, then Add.
- Description and Expiry can be set according to the requirement
Copy the new client secret value. You won’t be able to retrieve it after you perform another operation or leave this blade.
Again go to Resource Groups select Bot Channel and click on Overview. Copy the Application (Client) ID
Step 4: Connect bot to Azure
- Go back to SAP CAI.
- Open your bot, and go to the Connect tab
Again come back to AZURE PORTAL click on Resouce Groups, Select the Bot channel Click Configuration, and paste Messaging endpoint here and click on APPLY
Step 5: Test bot in Azure Web Chat
Go to Test in Web Chat, here you can test your bot by sending a message as shown below
Step 6: Test bot in Teams
It’s time to make a connection with MS Teams, Click on Channels select MS Teams
As shown below, click on Microsoft Teams Commercial and SAVE it.
A connection has been made. You can see Microsoft Teams connected and in a running state.
Click on Microsoft Teams and test it by sending the message.
How to Deploy SAP AI Chatbot on Telegram?
Start Chatting with BotFather by clicking START/ RESTART.
Done.
Click on “t.me/bot name” link on the telegram that has come with a congratulation message, It will take you to the chat.
Conclusion
And that’s it! Hope you’ve enjoyed building the Chatbot and its deployment on different platforms.
Thanks for sharing your knowledge... Wow!
Thank you for sharing this great use case with the community 🙂
I just want to let you know that you can submit this article for the SAP Conversational AI Tutorial Challenge if you want to, the guidelines and prizes are explained here.
Let me know if you have any questions!
Regards,
Paul
Thanks, Paul.
I will submit.
Anmol Sharma The MS teams at present does not handle utterances and replies back with a GUID ( key ID code).
Did you face this issue and the workaround since I see in your snapshot that your MS teams conversation seems to flow fine?
See my snapshot below