Calory tracker & meal logger with Telegram, Gemini AI and data tables

This can be your Alternative to Cal AI App Overview: A comprehensive n8n workflow demonstrating advanced AI agent orchestration, stateful conversation management, and multi-modal input processing for nutrition tracking applications. Technical Architecture: This workflow showcases production-ready patterns for building complex conversational AI systems in n8n: 1. Multi-Agent System Design Router Agent: Analyzes user intent and routes to specialized sub-agents Registration Agent: Handles new user onboarding with guided data collection Meal Logging Agent: Processes text, image, and voice inputs for nutrition analysis Update Agent: Manages multi-turn conversations for meal corrections Report Agent: Generates daily nutrition summaries Profile Agent: Handles user preference updates 2. Stateful Conversation Management The workflow implements a state machine using n8n Data Tables: Tracks conversation context across messages Enables multi-step workflows (e.g., "get meal ID → ask for new description → update") Prevents context loss in long conversations Auto-resets state after task completion 3. Multi-Modal Input Processing Text: Direct AI analysis using Google Gemini Images: Gemini Vision API with specialized nutrition analysis prompts Voice: Audio-to-text transcription followed by analysis Unified output format regardless of input type 4. Advanced Prompt Engineering Specialized system prompts for each agent role Structured output formatting for reliable data extraction Regional cuisine knowledge (African/West African foods) Error handling and clarification requests built into prompts 5. Data Persistence Layer Uses n8n Data Tables for: User profiles (calories, protein targets, preferences) Meal logs (descriptions, macros, timestamps) Conversation state (current step, context variables) 6. Telegram Integration MarkdownV2 formatting with auto-chunking for long messages Handles all Telegram message types (text, photo, voice) "Typing" indicators for better UX Learning Outcomes: By studying this workflow, you'll understand: How to build agent hierarchies with specialized roles State management patterns for complex conversations Multi-modal AI input processing techniques Production data persistence strategies in n8n Scalable webhook-based bot architecture Technical Requirements: n8n version 1.0+ (uses Data Tables feature) Google Gemini API access Telegram Bot Token Basic understanding of: AI agents, webhook triggers, data tables Difficulty Level: Intermediate to Advanced Use Cases Beyond Nutrition: The patterns here apply to any domain requiring: Multi-step user onboarding Data collection through conversation Record updates via natural language Report generation Profile management Components: 50+ nodes demonstrating n8n AI features 6 specialized AI agents 3 data tables with structured schemas Custom JavaScript for MIME type handling and markdown formatting Integration with Google Gemini (text + vision + audio models) What Makes This Workflow Unique: Most n8n AI examples show simple Q&A bots. This demonstrates enterprise-grade conversation management with real data persistence, making it suitable for actual production deployment. SETUP GUIDE Need Help? Reach me: https://www.linkedin.com/in/gerald-akhidenor-1ab1a45/ Work with me: https://dominixai.com/ My website: https://jobmonkey.dev My email: denorgerald@gmail.com Cal AI Nutrition Bot - Complete Setup Guide Who This Guide Is For: This workflow is designed for beginner to intermediate no-code builders who want to launch a nutrition tracking bot. You should be comfortable: Creating accounts on platforms (n8n, Google Cloud, Telegram) Following step-by-step instructions Copy-pasting API keys Basic JSON importing No programming experience required. If you can use Zapier or Make.com, you can set this up. Prerequisites (30 minutes to gather) 1. n8n Account Option A (Easiest): Sign up for n8n Cloud at n8n.io (14-day free trial, then $20/month) Option B (Free): Self-host n8n on DigitalOcean/Railway/