[key points]

Key points

• •AI agents' capabilities in legal and corporate analysis tasks have significantly increased, with Anthropic's Opus 4.6 model achieving nearly 30% in single attempts and 45% with multiple trials.
• •This significant improvement, partly due to new agentic features like 'agent swarms', indicates that progress in foundation models is not slowing down.
• •Although lawyers are not immediately threatened with replacement, they should be less confident than before about the future impact of AI on their profession.