AI Chat

The AI chat endpoint provides a streaming conversation interface powered by X.AI (Grok). The AI assistant can execute thogit and tag operations as tools during the conversation, allowing natural-language interactions with your knowledge base.

POST /chat

Send a message and receive a streaming AI response via Server-Sent Events (SSE).

Auth required: Yes

Request body:

Field	Type	Required	Description
`message`	string	Yes	The user’s message
`chat_id`	string	No	Chat session ULID to continue an existing conversation
`previous_response_id`	string	No	Response ID from a previous turn for conversation continuity

curl -X POST https://app.thogits.com/api/chat \
  -H "Content-Type: application/json" \
  -b cookies.txt -c cookies.txt \
  -N \
  -d '{"message": "Create a thogit called Meeting Notes and tag it with Project"}'

The -N flag disables output buffering, which is important for streaming responses.

Starting a New Chat

Omit chat_id to start a new conversation. The server will create a chat session and return its ID in a chat_session event.

Continuing an Existing Chat

Pass a chat_id and optionally previous_response_id to continue a conversation with full history context.

curl -X POST https://app.thogits.com/api/chat \
  -H "Content-Type: application/json" \
  -b cookies.txt -c cookies.txt \
  -N \
  -d '{
    "message": "Now add a due_date field set to next Friday",
    "chat_id": "01HQ9C3D4EFGHIJKL5MNOPQ6R",
    "previous_response_id": "resp_abc123"
  }'

SSE Event Types

The response is a stream of Server-Sent Events. Each event has a data field containing a JSON object with a type discriminator.

text_delta

A chunk of the assistant’s text response. Concatenate these to build the full message.

{"type": "text_delta", "content": "I'll create that thogit"}

tool_call

The assistant is executing a tool (e.g., creating a thogit or tag). Sent at both the start and completion of tool execution.

{"type": "tool_call", "name": "create_thogit", "status": "running", "result": null}

{"type": "tool_call", "name": "create_thogit", "status": "completed", "result": {"thogit_id": "01HQ4M8R3ZBNWY6K9J2X5V7Q0A"}}

done

The response is complete. Includes a response_id to use as previous_response_id in the next turn.

{"type": "done", "response_id": "resp_abc123"}

chat_session

Sent when a new chat session is created (when chat_id was not provided).

{"type": "chat_session", "chat_id": "01HQ9C3D4EFGHIJKL5MNOPQ6R"}

react_render

A dynamically rendered UI component from the assistant.

{"type": "react_render", "id": "component_1", "js": "...", "css": "..."}

error

An error occurred during processing.

{"type": "error", "message": "XAI_API_KEY not configured"}

Tool Execution

The AI assistant has access to tools that mirror the Thogits API:

Creating, updating, and deleting thogits
Creating and managing tags
Applying tags to thogits and setting field values
Searching thogits and tags
Querying the audit log

The assistant can chain multiple tool calls in a single response. The server enforces a maximum of 200 tool loops per request to prevent runaway execution.

Client Example

const response = await fetch('/api/chat', {
  method: 'POST',
  credentials: 'include',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    message: 'List all my thogits tagged with Project',
    chat_id: existingChatId,
  }),
});

const reader = response.body.getReader();
const decoder = new TextDecoder();
let buffer = '';

while (true) {
  const { done, value } = await reader.read();
  if (done) break;

  buffer += decoder.decode(value, { stream: true });
  const lines = buffer.split('\n');
  buffer = lines.pop(); // keep incomplete line in buffer

  for (const line of lines) {
    if (line.startsWith('data: ')) {
      const event = JSON.parse(line.slice(6));
      switch (event.type) {
        case 'text_delta':
          process.stdout.write(event.content);
          break;
        case 'tool_call':
          console.log(`Tool: ${event.name} [${event.status}]`);
          break;
        case 'done':
          console.log(`\nDone. Response ID: ${event.response_id}`);
          break;
        case 'error':
          console.error(`Error: ${event.message}`);
          break;
      }
    }
  }
}