Bot studio
The Bot studio is where you can create, manage, and test your AI chatbot. You can add knowledge sources, function calls, and response workflows to train your AI chatbot.
Creating a new AI chatbot
Follow the steps below to create a new AI chatbot in the Bot studio.
Step 1 Go to Bot studio
-
Log in to your Sendbird Dashboard.
-
Go to your Application > AI chatbot and click on Create bot + button in the top-right corner of the screen.
Step 2 Fill in the General tab
Start by filling in the general information of your chatbot in the General tab. The tables below list descriptions of each field in the Bot studio.
Bot profile
| Field | Description |
|---|---|
Profile image | (Optional) You can upload a profile image for your bot. IT will appear in the chatbot widget. |
Bot name | Give your bot a name. It will appear on the top of the widget as the bot's name. |
Bot ID | Pre-determined by Sendbird. You can copy the bot ID here and use it when necessary. |
Feedback | When turned on, you can collect users’ satisfaction with your bot’s responses. Displays thumbs up, thumbs down icons. |
Bot type | (Optional) Categorize your bot into types for organization and management. |
Mention bot for response | When turned on, your bot reacts only to direct mentions or commands starting with '/'. Ideal for privacy, this feature is auto-enabled in channels with 3 or more members to prevent unnecessary bot interruptions. |
Chat starters & replies
| Field | Description |
|---|---|
Welcome messages | Craft a friendly and inviting message that users will first see when they interact with your bot. This is your bot’s chance to make a good first impression, so consider what message will make users feel welcomed and eager to engage. |
Suggested reply | Propose some pre-defined replies that users can choose from when responding to your bot. This helps streamline the conversation and makes it easier for users to interact, especially if they’re unsure of what to say. |
AI Engine & instructions
| Field | Description |
|---|---|
AI Engine | Select the AI engine that powers your bot. See the list of available engines here. |
Instructions | (Optional) Define the main task or behavior of your bot here. Explain in simple terms what you expect the bot to do when interacting with users, such as answering specific types of questions or providing helpful information on a topic. You can also utilize the general template provided by Sendbird. |
Streaming response | If turned on, the words of your bot responses will be displayed as they are being generated. If turned off, the response will appear all at once only after it’s been fully generated. |
Temperature | Controls the randomness or creativity in the generated responses. A higher temperature value will result in more diverse and unpredictable responses, while a lower temperature value will produce more conservative and predictable responses. Acceptable values range from |
Maximum tokens | The maximum number of tokens to generate in the response by AI chatbots. One token generates roughly four characters for normal English text. Acceptable values range from |
Top P | A value that determines the diversity of response generated by AI chatbots via nucleus sampling. Higher values can lead to a variety of responses, but less reliable answers. |
Frequency penalty | A value between |
Presence penalty | A value between |
Style
You can customize the style and appearance of the chatbot widget. You can change the theme, primary color, bot message color, and widget icon. You can also turn on or off the Widget auto-open setting which automatically opens the chatbot widget when a user visits your website. Note that this setting does not apply in mobile environments and the preview.
Step 3 Fil in the Knowledge sources tab
Add sources to train your AI chatbot in the Knowledge sources tab.
Note: Make sure you have sources added to your knowledge center before connecting them to your chatbot. See Adding a knowledge source for more information.
- Link Knowledge sources to your chatbot
In the Bot Studio, under the Knowledge source tab, select the sources you want to connect to your chatbot. The bot will only respond with information from these sources.
This is what it should look like when the selected knowledge sources are successfully connected.
- Set up Advanced settings
Then, you can further set up advanced settings for your knowledge sources.
- Weekly auto-update: Available for Pro and Enterprise plans. This feature automatically refetches data from connected URLs each week to ensure your knowledge base is always up-to-date.
- Additional instruction for knowledge base: Customize the system prompts to better define the chatbot's responsibilities, responses, and limitations. This helps ensure your chatbot operates effectively within its intended scope. You can click on Show default prompt to see the original system prompts.
Step 4 Connect functions
Select function calls to connect to your chatbot under the Function calls tab.
Note: Make sure you have functions created before connecting them to your chatbot. See How to create a function for more information.
Step 5 Connect response workflows
Select response workflows to connect to your chatbot under the Workflow tab.
Note: Make sure you have response workflows created before connecting them to your chatbot. See Set up a response workflow for more information.
Step 6 Test your chatbot
You can test your chatbot in the Bot tester on the right.
Step 7 Save
Once you're all set, click Apply changes to save your settings.
Step 8 Add to your website
Add your AI chatbot to your website by clicking on the Add to my website button on the top-right corner of the screen.
The following are available integrations: