Boris Cherny, the creator of Claude Code, posted a screenshot on X showing his "orchestration": five independent agents running simultaneously in iTerm2. One agent writes tests, another refactors code, a third generates documentation, and the remaining two maintain web instances and route sessions between the browser and the local machine. Cherny says that this set of agents can replace a typical small engineering team.
The advantage does not lie in the raw power of Opus 4.5 but in its "thinking": depth of inference matters more than microsecond latency. Cherny calculated that a single developer equipped with five agents could dramatically lower project costs.
Cherny's public "manifesto" has sparked interest from investors and corporate clients. Anthropic is promising growth in enterprise subscriptions, while OpenAI, Google and Meta have already announced plans to improve their own coding agents, creating competitive pressure in the market.
Early adopters—large financial and technology firms—report faster feature delivery and reduced hiring expenses. However, agent stability currently depends on model updates, and a failure of any single agent can cause downtime.
The bottom line for business: deploy Claude Code now to cut costs and gain an advantage before competitors catch up.
Why this matters: A single developer with five AI agents can replace several engineers, shrinking payroll and accelerating product cycles. Implementing Claude Code today positions your firm ahead of the emerging wave of AI‑augmented development.