Single agent vs multi-agent: when subagents are worth it
Anthropic's Applied AI team published guidance arguing that enterprise agent architectures should default to a single agent loop with skills and tools loaded on demand, reserving subagents only for cases needing isolated…