Purple Chat Builder

Follow our best practices when scoping your use cases. This will set you up for success in your overall project plan

👔 Need help? Conversation Design Request

Here, you can build your own Purple chat mocks. Check out our idea library for some examples of what you can build.

Mocks

(566) Mock 1

Chat Messages

Edit as JSON

Quick Add

Full Templates

View all design patterns

Chat Settings

User Profiles

Bot

Bot Name

The name of your bot in mocks

Image from file

This will be set as your default bot image for all Purple Chats.
Warning: this will only be saved in your browser session and won't be included in the URL. Please use Image URLs for shareability.

Image from URL

This will be set as your default bot image for all Purple Chats & shareable

Gwen11:43 AM

I am WFH now and missing some laptop accessories to stay productive. It'd be great if I can get an external keyboard.

Searches ITSM for applicable resources

Moveworks11:43 AM

I found related request forms. Files the relevant one to provide important details for your agent.

Hardware request

Fill out this form to request any pieces of hardware you would like, including headphones, keyboards, and cables.

Yes

Get Help

Cancel

Build Purple Chats with ChatGPT

If your organization allows access to ChatGPT, you can use the following prompt to quickly generate a purple chat mock.

I have the following Typescript interfaces

enum SenderType {
    USER = "USER",
    BOT = "BOT",
    ANNOTATION = "ANNOTATION",
}

enum ButtonStyle {
    PRIMARY = "PRIMARY",
}

interface ChatMessage {
    from: SenderType;
    text: string;
    cards?: ChatCard[];
}

interface ChatCard {
    title?: string;
    text?: string;
    buttons?: ChatButton[];
}

interface ChatButton {
    style?: ButtonStyle;
    text: string;
}

I need to output an array of ChatMessages that represents a conversation. Here is an example.

[
  {
    "from": "USER",
    "text": "I am WFH now and missing some laptop accessories to stay productive. It'd be great if I can get an external keyboard."
  },
  {
    "from": "ANNOTATION",
    "text": "Searches ITSM for applicable resources"
  },
  {
    "from": "BOT",
    "text": "I found related request forms. Files the relevant one to provide important details for your agent.",
    "cards": [
      {
        "title": "Hardware request",
        "text": "Fill out this form to request any pieces of hardware you would like, including headphones, keyboards, and cables."
      },
      {
        "buttons": [
          {
            "style": "PRIMARY",
            "text": "Yes"
          },
          {
            "text": "Get Help"
          },
          {
            "text": "Cancel"
          }
        ]
      }
    ]
  }
]

You may also use the following markup tags for all "text" attributes. 
- Bold <b>Bold</b>
- Italics <i>Italics</i>
- Links <a href="https://moveworks.com"> Links </a>
- Emoji Unicodes 👍
- Break tags <br>

ONLY use Annotations to indicate system API behavior. The Annotations MUST follow the following format.

If the annotation is the first message:
"""
Trigger: Daily job (Hosted by iPaaS)
1.Get a list of live courses coming up
2. Lookup who is signed up for that course
3. Send the following Event to those users
"""

If the annotation is in the middle of the request.
"""
Moveworks extracts "PLS" passes to API:  

Inbound Request to System/Middleware: {"id": "PLS"}
System: Google Sheets

Process:
1. Establish a connection to this Google Sheet
2. Filter the data in the sheet where "Acronym" column matches "id" of the request
3. Return the data in the SINGLE row pertaining to the ID of the request (aka. the Acronym)

Outbound Response:
{"Acronym": "PLS", "Word": "Product Led Sales", "Meaning": "A sales approach where the product itself drives customer acquisition and expansion."}
"""

Make sure to use cards & buttons for any call to actions. Use card titles for bot questions, use buttons if there are any questions with discrete (multiple choice) options.

Please produce a conversation that models the following requirements. Your output MUST be written in strict, well-formed JSON. Tone: conversational, spartan, use less corporate jargon.

The chat starts when (a user asks... | a script detects...) << Continue the conversation >>