Skip to content

Latest commit

 

History

History
21 lines (11 loc) · 842 Bytes

File metadata and controls

21 lines (11 loc) · 842 Bytes

Developer Interaction Patterns with Proactive AI: A Five-Day Field Study

Why It Matters

This field study grounds proactive coding-assistant design in real developer behavior rather than lab-only task completion.

Proactivity Signal

The agent proactively surfaces code-quality suggestions, and the study analyzes when developers accept, ignore, or experience those suggestions as disruption.

Evaluation Setup

The paper observes developers over five days in an IDE setting and analyzes interaction timing, workflow phase, acceptance, and perceived value.

Key Limitations

The deployment period is still short, and the study focuses on coding rather than broader computer-use workflows.

Use For

Use this for evidence about workflow boundaries, timing calibration, and real-world acceptance of proactive developer tools.