Overview

• Users can activate the new Learning preset in Claude.ai’s style dropdown to receive Socratic-guided interactions that prompt self-discovery.
• After updating Claude Code, developers run “/output-styles” to choose between Default, Explanatory or Learning modes within their coding workflows.
• In Explanatory mode Claude narrates its decision-making process while in Learning mode it pauses with “#TODO” markers asking users to write code snippets.
• All learning behaviors run on adjustable system prompts rather than model fine-tuning, enabling Anthropic to iterate features rapidly based on user feedback.
• The rollout aligns with back-to-school launches of OpenAI’s Study Mode and Google’s Guided Learning as AI providers compete for educational use cases.