🏢 Business / Industry: Content creation / Social media management

🎯 Main Goal: Automate content generation, image formatting, and publishing to Telegram through a single AI-powered conversational interface.

🔹 Context

Content creators and social media managers spend significant time crafting posts, selecting image formats for different platforms, and manually approving each piece before publishing. This slows down content pipelines and increases the risk of human error or inconsistency across channels.

🔹 Problem

Creating and publishing content to a Telegram channel required manual effort at every step — writing copy, generating visuals, choosing the right format per platform, and coordinating approval before posting. Without automation, the process was slow, repetitive, and hard to scale.

🔹 Solution

An AI-powered Telegram bot was built to handle the full content generation and publishing pipeline — from understanding the user’s request to generating an image in the correct format, collecting approval, and automatically posting to a Telegram channel — all through a single conversational interface.

🔹 Process

• Trigger & input handling: The Telegram Trigger node listens for incoming user messages. A prepare_data step normalizes the input, and a prompt node formats it for downstream processing.
• Bot typing indicator: A sendChatAction message signals to the user that the bot is processing, improving perceived responsiveness.
• Intent routing: A Switch node analyzes the message and routes it across multiple branches — Welcome message, Image generation (Instagram, Facebook, or fallback), or a Help/How-to-use response — based on detected intent and platform keywords.
• Welcome flow: If the message is a greeting or first interaction, a Claude model generates a personalized welcome response, which is sent back to the user via Telegram.
• Image generation: Depending on whether the user’s message contains insta, fb, or neither, the appropriate image generation node is triggered with the correct aspect ratio and format. Results are merged and passed forward.
• Caption generation: A Claude model generates a platform-appropriate caption for the image, which is sent to the user alongside the visual preview.