Currently, the AI draft feature appears to generate responses based solely on the active email’s content. This limits contextual accuracy and produces replies that can feel disconnected from the established communication tone or relationship history.

A key enhancement would be enabling context-aware generation by incorporating prior message data between the sender and recipient. Specifically:

Use historical correspondence with that contact (or within the same thread) to infer tone, formality level, and conversational cadence.

Apply lightweight sentiment or style analysis across prior exchanges to calibrate response tone (e.g., formal vs. casual, concise vs. detailed).

Optionally expose a toggle or “context depth” setting so users can control how much past context informs the draft.

This would make the drafts system relationship-aware, improving coherence, personalization, and alignment with ongoing communication patterns.