THE ACTORS
SECTION 02
ISSUE 001
Agent Teams Coordinate in Parallel
Observed now: Claude Code's agent-team preview lets multiple agents work in parallel and coordinate autonomously on divisible tasks, while RE-Bench evaluates agent designs on time-bounded machine-learning research engineering work. Together they establish coordinated multi-agent work as a testable system pattern without claiming mature org charts, managers, or governance.
Why this idea is here
What the evidence establishes.
Observed evidence: Anthropic documents a Claude Code research preview in which multiple agents work in parallel and coordinate autonomously; RE-Bench compares frontier agents and humans on time-bounded machine-learning research engineering environments with varying agent designs. These sources support coordinated teams as an evaluable pattern, not mature organizational structure.
Source ledger
Read the sources.
- S01Introducing Claude Opus 4.8
official model release / published 2026-05-28 / retrieved 2026-07-10
- S02RE-Bench
peer-reviewed conference paper / published 2025-07-01 / retrieved 2026-07-09
- S03Task-completion time horizons of frontier AI models
current independent evaluation tracker / published 2026-05-08 / retrieved 2026-07-10