Deep Research Team

This spawns a team of independent researchers (2-4 depending on scope) who work in rounds, with a lead agent triaging findings and dispatching follow-up tasks based on what emerges. The architecture is thoughtful: researchers never see each other's work to avoid groupthink, and in comprehensive mode one researcher explicitly verifies another's single-source claims. Scales from focused two-researcher jobs to four-round deep dives with cross-verification. The process persists state to disk so you can resume interrupted research. Skip it for simple lookups or anything answerable in one or two searches. It's built for questions where the first round of findings tells you what to investigate next.