Sample Chats
Example conversations you can drop into chat UIs, tests, or mocks. Pick a format tab to copy structured data, or use the preview to eyeball layout.
[
{
"role": "user",
"content": "I've been arguing a lot with my partner lately and I don't know how to fix it.",
"timestamp": "2026-04-05T21:00:00Z"
},
{
"role": "assistant",
"content": "Frequent arguments often come from miscommunication or unmet expectations. Have you talked openly about what's actually causing the tension?",
"timestamp": "2026-04-05T21:00:05Z"
},
{
"role": "user",
"content": "We try, but it usually turns into another argument.",
"timestamp": "2026-04-05T21:00:12Z"
},
{
"role": "assistant",
"content": "In that case, it might help to change how you communicate. Try focusing on how you feel instead of what the other person is doing wrong.",
"timestamp": "2026-04-05T21:00:16Z"
},
{
"role": "user",
"content": "Can you give an example?",
"timestamp": "2026-04-05T21:00:22Z"
},
{
"role": "assistant",
"content": "Instead of saying 'You never listen to me', try 'I feel unheard when I try to explain things'. It reduces defensiveness.",
"timestamp": "2026-04-05T21:00:26Z"
},
{
"role": "user",
"content": "That makes sense. Anything else?",
"timestamp": "2026-04-05T21:00:32Z"
},
{
"role": "assistant",
"content": "Yes—timing matters. Avoid difficult conversations when emotions are already high. Also, actively listen instead of preparing your response.",
"timestamp": "2026-04-05T21:00:36Z"
}
]Fields
role
Defines the sender of a message in the chat. user is a user message, assistant is an AI-generated message, and system is optional system-level instructions.
content
The actual message text. Can include plain text, markdown, or formatted content depending on your UI.
timestamp
ISO 8601 formatted date string (e.g. "2026-04-05T10:00:00Z").
Extensions
Depending on your use case, you can extend the format with additional metadata:
type ExtendedChatMessage = {
role: "user" | "assistant" | "system";
content: string;
timestamp: string;
// optional
id?: string;
name?: string;
delay?: number; // simulate typing latency (ms)x
usage?: {
promptTokens?: number;
completionTokens?: number;
totalTokens?: number;
};
model?: string; // e.g. "gpt-4.1"
};