Anthropic says one AI agent beats many subagents now that models can handle complexity alone

Anthropic is advising companies to replace complex architectures of multiple specialized subagents with single, more capable models. The company reports that using a single agent can double execution speed while maintaining performance on complex tasks. To support this shift, Anthropic has released a technical migration kit on GitHub designed to help organizations consolidate their workflows. This change aims to simplify development by leveraging the growing ability of models to manage large amounts of context independently.