AI Systems Atlas

G-02Single Agent vs Multi-Agent

/coding-agents/single-vs-multi

One agent with full context and no coordination — or several specialists working in parallel, plus the overhead of keeping them coherent.

Step-by-step walkthrough2 variants · 8 steps

One agent

  1. One agent, one context: everything it learned exploring is still there while editing.
  2. Serial, so wall-clock bound — but every step sees full history.
  3. Done. Slower, simpler, often more correct.

Multi-agent

  1. The orchestrator splits the feature across four specialists.
  2. They run in parallel — each in its own context, blind to the others.
  3. The hidden tax: agents must sync assumptions they no longer share.
  4. Integration reconciles four partial views — real merge conflicts, real re-work.
  5. Faster wall-clock only if the task splits cleanly. More agents ≠ better.
Key insight

Multi-agent buys wall-clock speed on cleanly-splittable work and nothing else. Coordination cost grows with coupling, and most features are more coupled than they look.

Failure mode

Splitting a coupled task: four agents each holding a quarter of the picture will confidently build four incompatible quarters.

Further reading
RelatedHierarchicalParallel swarmWorktrees

Elsewhere in the atlas

Harnesses →Security →Evals →Context →Tools →Ops →