As a journalist, I built five specialized AI assistants for various tasks like fact-checking, brainstorming, and editing. However, this led to chaos and decision fatigue. So, I created one “routing” agent that decides which assistant should handle each task. This routing agent has become the most useful in my workflow.
The idea is simple: just like a productive team has specialists, AI models can too. My routing agent works by evaluating tasks using five questions: What is the user’s goal? Which specialist is best suited to this task? Should more than one specialist be involved? In what order should they work? Is any information missing before work begins?
To create your own routing agent, start by identifying common tasks you repeat. Create a Custom GPT for each job and build a routing agent using the prompt: “You are my AI Routing Agent…” The routing agent learns about new specialists and incorporates them into future workflows.
This concept is similar to Sakana’s AI assistant but keeps agents within the same platform. It works well with ChatGPT agents and can be adapted for various tasks, like image generation or social media posts. By simplifying your AI workflow, you’ll spend less time managing tools and more time getting work done.
Try building a routing agent yourself and let me know what you think in the comments!
Source: https://www.tomsguide.com/ai/i-built-five-chatgpt-agents-then-replaced-them-with-one-master-ai