Wikipedia-gebaseerde Q&A AI-assistent
Een AI-assistent die gebruikersvragen beantwoordt met feitelijke en goed onderbouwde informatie, gebruikmakend van de RIG-methode om antwoorden te baseren op Wi...
4 min lezen
Flows are a versatile tool that allows you to connect many variations of components. That’s why you need a way to tell the bot you’re done creating, and it’s time for it to craft an answer based on the connected components.
The Chat Output component is how you tell the chatbot to start crafting an answer based on the connected components. One flow can have several outputs coming together in a single reply. For example, one output generates a text answer from AI, while a second one adds relevant links from your knowledge base.
Always being at the end of the chain, it doesn’t connect to anything, but any component that outputs text can connect to it. To see the options, just hover over the connector handle.
Inputs
The Chat Output component accepts the following inputs:
| Input Name | Type | Required | Description |
|---|---|---|---|
| Text | Message (str) | Yes | The main message text to be displayed in the chat output. |
| Images | Attachment | No | Images to be shown alongside the message in the output. |
| Data Template | str | No | (Advanced) Template for converting data to text. If left empty, defaults to the data’s ’text’ key. Useful for custom formatting. |
Input Details
- Text: The core content that will appear as a chat message. This is a required field and must be provided for the component to function.
- Images: Optional attachments that can be displayed with the chat message, enhancing the output with visual elements.
- Data Template: An advanced option for formatting the message output. You can use placeholders to customize how data is rendered as text.
Outputs
The component produces the following output:
| Output Name | Type | Description |
|---|---|---|
| Message | Message | The processed chat message (including images if provided), ready to be displayed in the UI. |
- The output is typically used as the final step in a workflow, making the result visible to users.
How to use multiple outputs
Each flow can only contain a single Chat Input component but may contain various outputs. These are useful for creating complex answers consisting of several elements. Let’s look at an example of a flow with several outputs.
Imagine the user wants to know what an IVR is. Let’s say you want your bot to answer the question in a lighthearted, conversational way and then link further reading from your website.
To achieve this, you will want to create two separate outputs:
- The first output connects the Generator component to create a text answer with ChatGPT’s help.
- The second output retrieves knowledge from your knowledge sources and displays it in a visual, interactive way thanks to the URL Widget:
The final answer will combine these two types of outputs:
Separated outputs provide greater flexibility and control over the chatbot. By adding other outputs, you can continue making the answers more complex. For example, you can make the related sources more elaborate by adding Google Search to the mix, or you can introduce follow-up questions to prompt the user to ask more questions.
