Description
Intelligent Telegram chatbot with DeepSeek AI & Google Docs memory โ context-aware and personalized
๐ Stop robotic replies โ this AI agent uses DeepSeek models to remember past chats, understand context, and provide tailored responses over time
๐ง What Problem Does It Solve?
Most AI chatbots forget everything after each session. This workflow solves that by adding memory, context awareness, and multi-model capabilities using DeepSeek's latest reasoning and conversational engines.
It gives you a smarter chatbot that can:
๐น Recall previous chats
๐น Respond contextually
๐น Understand intent over time
๐น Scale across message types (text, audio, image)
โ๏ธ How It Works
๐ค Listens to Telegram messages using a webhook
๐งพ Validates user identity before processing
๐ Routes messages by type (text, audio, image)
๐ช Sends input to DeepSeek-V3 or DeepSeek-R1 based on use case
๐ Retrieves relevant memories from Google Docs
๐ Saves new memories after every meaningful interaction
๐ฌ Sends structured responses back via Telegram
โจ Key Features
๐ก Telegram-based interface for easy chat access
๐ง AI responses powered by DeepSeek-V3 (chat) and DeepSeek-R1 (reasoning)
๐๏ธ Google Docs used as long-term memory storage
๐ Memory-aware system prompt guides the agent behavior
๐ Maintains conversational context with session window buffer
โ Built-in error messaging and user validation
๐ธ Supports multiple media types: text, audio, and images
๐งฐ What You Need
โ
Telegram bot token and webhook URL
โ
DeepSeek API key (free or paid)
โ
Google Docs API access for storing memories
โ
n8n hosted or self-hosted instance
๐ ๏ธ Setup Instructions
๐ Connect the workflow to your Telegram webhook
๐ Add credentials for DeepSeek API and Google Docs
โ๏ธ Choose between DeepSeek-V3 and DeepSeek-R1 for different AI tasks
๐ง Use the system prompt to define agent tone, memory rules, and fallback behavior
๐ Set a dedicated Google Doc for memory storage
๐งช Test with one user before scaling to groups
๐ Integrations
Telegram
DeepSeek AI
Google Docs
n8n AI Agent (LangChain wrapper)













