The most common feedback we received from editors using the previous review interface was simple: “I want to see what changed without losing sight of what was there.” For a heavy stylistic edit, the diff view alone is noisy — every other word seems marked — and the underlying prose disappears behind the markings. The new split-view editor puts both back where they belong.
What’s new
Original on the left, edited on the right. The two panes stay aligned as you scroll, so the paragraph you’re reading on one side is always next to its counterpart on the other. Click any sentence in one pane and the corresponding location in the other pane is highlighted.
Two distinct change layers — AI and editor. Every change is attributed: AI revisions are shown in one colour, your own revisions in another. When you export to Word, both layers ship as track changes with their original author names, so a downstream reviewer can see at a glance who suggested what.
Each change — here making verb tense consistent — appears as a separate tracked change with an explanation, ready to accept, reject, or adjust.
Floating AI-menu for criterion-driven edits. Hover over any sentence or selection and the floating AI-menu appears, letting you ask for a targeted rewrite against a specific criterion. The result lands as a tracked change you can accept or reject. We’ve given the floating AI-menu its own write-up — see the floating AI-menu post for details on what it does and how to use it.
Why the layering matters
Authors don’t just want to know what changed — they want to know who changed it. The split editor enforces this distinction by keeping AI edits and your edits in separate layers and giving them different author names on export.
Track changes is only useful when it tells you a story. Who proposed what, in what order, and which decisions stuck — that story has to survive into the final file, or the whole transparency argument falls apart.
Updated AI models, too
While we were at it, we updated the AI models the editor can use. EditBook.ai now offers the latest generations from all three major providers: OpenAI GPT-5.5, Anthropic Claude Opus 4.7, and Google Gemini 3.1. Each model has slightly different strengths — Gemini for long-context analysis, Claude for nuanced stylistic editing, GPT for precise corrections — and you can pick per task. Model selection happens at the project level, so you can match the model to the work in front of you.
Where to find it
The split-view editor is the default when you open any reviewed manuscript. You can switch to the classic single-pane view from the editor’s top-right menu if you prefer it for very light edits.
It is, like everything in EditBook.ai, designed around the same principle that we keep coming back to: the AI does the heavy lifting, but the editor stays in charge of the page.
Posted by the EditBook.ai dev team. We write here when we ship something new.
