How to build a platform of agents

A walkthrough of building a platform of AI agents using llm.rb and ActiveRecord in Ruby. The pattern treats agents as ordinary ActiveRecord models with state, tools, skills, and user associations — rather than separate services. Two concrete examples are shown: a Daytrader agent and a StockInvestor agent, each belonging to a user, configured with domain-specific tools and skills, and capable of running concurrently via threads. The approach enables per-user personalized agents that integrate naturally into the application's data model.