Tenure is a mobile trust‑calibration layer designed to make agentic AI safer, more transparent, and more controllable.

Built in response to real incidents where AI agents caused irreversible damage, such as deleting production databases or overwriting entire directories, the project reframes the problem not as a failure of intelligence, but as a failure of the human, AI relationship. Drawing on two decades of HCI research on calibrated trust, Tenure introduces domain‑based permissions, a promotion‑and‑demotion “tenure” mechanic, and universal reversibility through Action Receipts. Instead of granting blanket autonomy, users can observe, adjust, and roll back AI actions across email, code, files, and more. The result is a system where trust is earned action by action, making AI collaboration visible, accountable, and safe, especially in a world where mobile devices are becoming the natural supervision layer for autonomous agents.