Telegram AI Assistant: Memory Summarization & Dynamic Routing

This n8n workflow creates a smart Telegram AI assistant that balances cost, speed, and precision. It uses PostgreSQL to store chat history in a single row per conversation, reducing database queries and speeding up responses by ~0.3s. AI-powered summarization keeps only relevant context, preventing token waste on irrelevant history.

Dynamic routing intelligently selects models: lightweight Gemini Flash Lite for simple queries and powerful Gemini Pro for complex tasks, slashing costs while maintaining answer quality. Voice message support converts audio to text for seamless interaction. Error-proof MarkdownV2 formatting ensures Telegram-compatible responses.

Ideal for businesses or individuals building affordable, responsive AI chatbots on Telegram. Use cases include customer support bots, personal assistants, or community managers needing context-aware, voice-enabled conversations without high API expenses.