> ## Documentation Index > Fetch the complete documentation index at: https://docs.sequencehq.com/llms.txt > Use this file to discover all available pages before exploring further. # Watchtower > A centralized command center to review, approve, and audit work completed by agents Watchtower Watchtower is a centralized command center where finance operators review and approve autonomous work, making AI actions visible, auditable, and controllable. ## Overview When AI agents work autonomously in the background, creating transparency around completed tasks is essential for trust. Watchtower surfaces all agent-completed work in a unified interface, giving you full visibility and control over what's happening. ## Key capabilities ### See what's completed and what needs review Watchtower displays all work completed by your agents through activity feeds. View `Needs review` to see which tasks require your attention. The number of tasks requiring your attention is also shown in the navigation. ### Full auditability Every action an agent takes is logged and traceable. Review what it did, what data it used, and what reasoning led to each outcome. ## Getting started Open Watchtower to immediately see: * What your agents have completed * Items requiring your review or approval * Full audit trails for any action * Exceptions that need manual attention For more context and principles behind Watchtower, read our blog post on [Building a human in the loop interface](https://www.sequencehq.com/blog/building-a-human-in-the-loop-interface).