Tabularia
Write and revise your whole book.
7+ months of sustained development to build and launch Tabularia.
Book-writing software that keeps the page, the structure, and every revision together in one connected manuscript, from first draft to finished work. Write close up while the outline and minimap keep the whole book navigable.
Follow a character arc, plot line, or recurring idea back to the passages that carry it. Try another version of a scene, chapter, or the whole manuscript, compare it with the current draft, and merge back only the changes you want. Review an editor’s changes and comments with the surrounding writing still in view.
A writing workflow that stays connected.
A long manuscript needs more than a text box. Writers need to move between a passage and the larger story, keep related ideas within reach, and understand which version they are working on. I built the writing surface, navigation, and revision tools around those needs, so the page and the structure remain connected as the book changes.
The timeline view brings related passages together across the manuscript. The review workflow keeps an author’s writing and an editor’s working copy connected, with changes that can be inspected before they are accepted. These are product decisions about how people work, backed by the storage and application behavior needed to make them reliable.
The engineering behind the writing.
My work spans the interface, manuscript model, revision history, local storage, cloud synchronization, and publishing tools. A writer’s changes need to survive navigation, interruptions, and revisions from another session. I work through those cases with automated tests and browser checks, including long manuscripts and mobile layouts.
AI assistants can use the connector to read the relevant writing and work with the manuscript through defined tools. The integration gives them source passages and locations, while the app keeps its own rules for revisions and review. I built the product and the surrounding workflow, using AI during development while remaining responsible for design decisions, code review, testing, and what reached production.
How I connected AI assistants to the writing
I built Tabularia’s MCP integration so assistants such as Claude Code and ChatGPT can search manuscript passages and retrieve context from Chapters, Scenes, and other Working Scopes. The integration uses bounded reading windows, paginated results, and links back to source passages, with explicit notices when results are incomplete.
