🤖 Cheaper, faster, accurate answers with memory summarization & dynamic routing!
⚡ 1,403 views · 🤖 AI Chatbots & Agents
Description
🤖💬 Smart Telegram AI Assistant with Memory Summarization & Dynamic Model Selection
> Optimize your AI workflows, cut costs, and get faster, more accurate answers.
📋 Description
Tired of expensive AI calls, slow responses, or bots that forget your context?
This Telegram AI Assistant template is designed to optimize cost, speed, and precision in your AI-powered conversations.
By combining PostgreSQL chat memory, AI summarization, and dynamic model selection, this workflow ensures you only pay for what you really need. Simple queries get routed to lightweight models, while complex requests automatically trigger more advanced ones. The result? Smarter context, lower costs, and better answers.
This template is perfect for anyone who wants to:
- ⚡ Save money by using cheaper models for easy tasks.
- 🧠 Keep context relevant with AI-powered summarization.
- ⏱️ Respond faster thanks to optimized chat memory storage.
- 💬 Deliver better answers directly inside Telegram.
✨ Key Benefits
- 💸 Cost Optimization: Automatically routes simple requests to Gemini Flash Lite and reserves Gemini Pro only for complex reasoning.
- 🧠 Smarter Context: Summarization ensures only the most relevant chat history is used.
- ⏱️ Faster Workflows: Storing user + agent messages in a single row reduces DB queries by half and saves ~0.3s per response.
- 🎤 Voice Message Support: Convert Telegram voice notes to text and reply intelligently.
- 🛡️ Error-Proof Formatting: Safe MarkdownV2 ensures Telegram-ready answers.
💼 Use Case
This template is for anyone who needs an AI chatbot on Telegram that balances cost, performance, and intelligence.
- Customer support teams can reduce expenses by using lightweight models for FAQs.
- Freelancers and consultants can offer faster AI-powered chats without losing context.
- Power users can handle voice + text seamlessly while keeping conversations memory-aware.
Whether you’re scaling a business or just want a smarter assistant, this workflow adapts to your needs and budget.
💬 Example Interactions
- Quick Q&A → Routed to Gemini Flash Lite for fast, low-cost answers.
- Complex problem-solving → Sent to Gemini Pro for in-depth reasoning.
- Voice messages → Automatically transcribed, summarized, and answered.
- Long conversations → Context is summarized, ensuring precise and efficient replies.
🔑 Required Credentials
- Telegram Bot API (Bot Token)
- PostgreSQL (Database connection)
- Google Gemini API (Flash Lite, Flash, Pro)
⚙️ Setup Instructions
- 🗄️ Create the PostgreSQL table (
chat_memory) from the Gray section SQL. - 🔌 Configure the Telegram Trigger with your bot token.
- 🤖 Connect your Gemini API credentials.
- 🗂️ Set up PostgreSQL nodes with your DB details.
- ▶️ Activate the workflow and start chatting with your AI-powered Telegram bot.
🏷 Tags
telegram ai-assistant chatbot postgresql
summarization memory gemini dynamic-routing
workflow-optimization cost-saving voice-to-text
🙏 Acknowledgement
A special thank you to Davide for the inspiration behind this template.
His work on the AI Orchestrator that dynamically selects models based on input type served as a foundational guide for this architecture.
💡 Need Assistance?
Want to customize this workflow for your business or project? Let’s connect:
📧 Email: johnsilva11031@gmail.com
🔗 LinkedIn: John Alejandro Silva Rodríguez
🔗 Nodes Used
Postgres, Telegram, Telegram Trigger, AI Agent, Basic LLM Chain, Structured Output Parser
📥 Import
Download workflow.json and import into n8n:
Workflow menu → Import from File