Actionbook Editor: A reliable WYSIWYG environment for authoring your AI’s rules

Operations teams need to update AI logic quickly, without depending on engineering for every change or worrying whether an edit will behave as expected.

The updated Actionbook Editor consolidates editing to a single WYSIWYG surface and pairing it with a real-time read-only Viewer panel so authors always see what the AI will see before saving.

How it works

• Single editing surface: One panel, one mode. Author and update actionbook logic directly with consistent, reliable rendering at any level of complexity.
• Toolbar-based block insertion: Add conditional logic, key points, note annotations, and formatted text directly from the toolbar.
• Real-time Viewer panel: The right panel updates live as you edit. Switch between the Source tab and the Steps tab to verify what the AI sees without saving.
• Click-to-navigate: Click any component in either Viewer tab and the editor scrolls directly to the corresponding section on the left. Instant orientation without leaving the editing surface.
• Built-in Tester: Test changes against your actual AI agent without leaving the screen.

Actionbook Editor is a part of Trust OS 2.0 — making it practical for operations teams to own and maintain AI behavior directly, without requiring engineering involvement for every change.