Claude 4 Arrives — and AI Starts Doing Whole Jobs

28 May 2025

Anthropic has launched Claude Opus 4 and Claude Sonnet 4, and made its Claude Code tool generally available. The benchmark charts say what benchmark charts always say — better, faster, cleverer — but the meaningful shift is in stamina: these models are built to work on a task for an extended stretch, using tools, checking results and correcting course.

From answers to outcomes

Eighteen months ago, an AI session meant a question and an answer. The new generation is designed for briefs: "research these five competitors and draft a comparison", "clean this spreadsheet and flag the anomalies", "build me a working prototype of this idea".

Developers were the first to feel it — hence the excitement around AI coding tools — but the same capability applies to marketing operations, reporting, research and content production.

A word of balance

Longer autonomy means longer opportunity to go confidently wrong. The businesses getting real value from this generation share three habits:

• They give the AI clear briefs with success criteria, exactly as they would a freelancer.
• They review outputs at checkpoints, not just at the end.
• They keep a human's name against every deliverable, because "the AI did it" has never once satisfied a client.

The tools are ready for real work. The management skills to direct them are now the scarce resource.