Viewer Suggestions
Viewers can't edit documents directly, but they have a powerful way to give feedback. Here's how it works:
Select a passage
Press record
Suggestion created
This creates a clean feedback loop: viewers give voice-based notes on specific passages, and owners decide what to incorporate โ without giving viewers direct edit access.
Accepting and Rejecting
Owners and Editors see suggestion cards appear in the document with the viewer's feedback and a proposed edit. For each suggestion, you can:
- Accept โ applies the suggested edit to your document
- Reject โ dismisses the suggestion without making changes
Each suggestion shows who made it, what text it targets, and the AI-processed edit instruction.
The Editor Agent
The Editor Agent runs automatically after edits. It checks for cross-document consistency across all documents in your project.
If your edit introduces a contradiction โ a character's name changes, a timeline doesn't line up, a location is described differently โ the agent identifies the affected documents and creates suggestions to fix the inconsistencies.
It focuses on:
- Character names and descriptions
- Timeline and chronology
- Locations and settings
- Plot points and story details
- World-building elements
AI agents never modify your document without permission. They create suggestions that you can accept or reject โ just like viewer suggestions.
The Fact-Checker
The Fact-Checker helps you verify factual claims in your document. Select the text you want to check, and the fact-checker will:
Extract claims
Search the web
Evaluate
Report
Each fact-check result comes with a verdict โ verified, unverified, or contradicted โ along with source links and snippets so you can verify the findings yourself.
This is especially useful for research documents, journalism, and any content where factual accuracy matters.
How Agents Appear
AI agents show up as collaborators in your project with their own emoji avatar. Their suggestions appear alongside viewer suggestions โ you accept or reject them the same way. Each agent run is tracked with its duration and cost, visible in the suggestion card.