How I'm thinking about multi-agent systems in n8n
Multi-agent systems can completely change your AI workflows. 1 orchestrator agent + specialized sub-agents = ideal setup in n8n. For example, instead of one agent juggling 25 tools w/ a massive prompt, you can create: -- An orchestrator that routes all the requests -- An email agent that handle all communication -- A content agent that creates and researches -- A calendar agent that manage scheduling And give your Main Agent access to all these sub-agents as specialized "tools". This way, each agent uses the ideal model for its specific task + the components are reusable across multiple workflows.