Principlev1
Never add agents to sequential reasoning tasks—multi-agent
Never add agents to sequential reasoning tasks—multi-agent coordination on sequential dependencies increases overhead without increasing throughput.
Why This Is a Principle
Derives from The mind operates as multiple semi-autonomous processes (multiple agents), The behavior of a system arises from its structure, not from (structure determines behavior), and The number of unique communication channels in a team of n (communication channels scale as n(n-1)/2). This is Brooks's Law applied to agentic systems: given that sequential tasks cannot be parallelized, adding agents only increases coordination cost. This is prescriptive and context-specific.