Harvey turns legal context into stronger drafts with GPT-6 Astra

The work applies GPT-6 Astra to legal drafting, with outputs described as more structured and context-aware legal documents. Relative to generic text generation, the emphasis is on structured document output oriented to legal context. The loaded text is a single summary sentence and provides no methods, samples, controls, or quantitative results.

The work positions lawyers as focusing on strategy while drafting is handled with model assistance. This reflects an arrangement of drafting and strategy roles within a legal workflow. Based on the summary phrase 'freeing lawyers to focus on strategy', which is a positioning statement rather than an empirical measurement.

The result is aimed at the specific setting of legal drafting, for practice teams that want model-assisted structured, context-aware drafts; the loaded text is summary-level and does not state applicable jurisdictions, document types, or deployment conditions.

A reader would still want to know how the model is used in legal context, how draft quality is measured, and whether the effect of 'freeing lawyers to focus on strategy' has verifiable support; because the loaded text is a fast parse without figures or data, these remain open questions for now.