How This Telegram AI Chatbot Works

This page explains in detail how the Grok bot works in Telegram: request handling, dialogue context, and service reliability.

Request flow

Telegram user message
        |
        v
Bot service (auth, quota, basic checks)
        |
        v
Grok API request
        |
        v
Bot service formats response
        |
        v
Telegram reply to user

What context is remembered

To keep conversations coherent, only short-lived context from the last two messages is used.
Long-term full chat history is not stored.
New Chat immediately resets the current session context.

What data is processed

Telegram user ID is used for quota, balance, and subscription checks.
Current request text and short context are used to generate a reply.
Technical metadata is used for request accounting and response delivery.

Chat behavior

Bot replies are sent as replies to your messages.
Typing indicator is shown while a response is being generated.
Responses support rich text formatting, including readable code blocks.

Available modes

Quick (default) is optimized for fast and affordable daily use.
Smart is for more complex prompts and deeper analysis.
Internet provides access to fresh web information.
Images generates images from text prompts and can analyze uploaded images.
Expert is an advanced mode for difficult tasks, detailed explanations, code work, and non-standard requests.

Latency and reliability

Response time depends on Grok API load and request complexity.
For transient issues, the bot retries request processing automatically.
In most cases, users do not need to resend the same request.