A few months ago I started writing a large, continuity-heavy story, and looking for software that could keep the world in its head. None of it could. I am building TexFaber to close that gap.
I ended up with hundreds of pages of notes. The number of tools I needed kept growing. Stitching them together was not keeping up.
01The start
I needed to track plotlines in detail: backstories, timelines, and my own arcs. I wanted to respect a complicated canon. I drafted in Google Docs and got to about 70,000 words before it became insufficient. My notes were scattered everywhere, and spreadsheets made it worse. That is when I started looking at paid tools.
02How I build things
I have tinkered with AI in my spare time for years. I have a background in TV and radio technical operations, so I tend to automate the boring bits.
In television I learned HTML, CSS and JavaScript well enough to make tools for the organisation. I built a navigation page for our intranet, and a news feed that scanned RSS and pulled related YouTube videos. That is mostly what writing software is: workflow and gluing systems together.
Since around 2023 I have used AI to write small Python scripts that let me skip the big editor for quick jobs: crop a clip, apply an effect.
Each tool I found does one job well and nothing else. You are expected to stitch them with time, patience, and another subscription. None of them could handle a project from start to finish.
03What I found
Grammarly polishes sentences that are not trying to be fiction. It will flag a deliberate fragment and push you toward bloodless prose. The rest of that category are proofreaders. They do not know the story.
Scrivener is the established option. It has a binder and a corkboard, so you can split a manuscript into hundreds of pieces and rearrange them. It is a one-time purchase, per platform, and major versions have been a paid upgrade. You write a scene onto a card, drag it, and that part of the book moves with it. It takes time to learn, the interface feels dated, there is no AI, and compiling a finished book is complicated.
NovelCrafter came closest to what I actually needed. The Codex is a wiki for characters, places, lore, and threads, and it feeds the model so help knows who people are. You have to bring your own API key. It is a good piece of software and it is actively developed, so anything I said about its limits would be out of date by the time you read this.
SudoWrite is AI-first: a chapter engine, fiction-tuned prose, a story bible. When it works, the sentences are good. Then the credit meter appears. You end up managing credits instead of chapters, and when they run out the chapter stops. Plots slip between those fills, and you still export into something else to finish the book.
World Anvil will hold a magic system and a trade route. Pins on a map open articles; articles send you back to the map. World Anvil does world building better than anyone, but it still misses out on a bunch of features needed for the best writing experience.
Then there are the general-purpose models: ChatGPT, Gemini, Claude, Grok. They are very good at some things but struggle with remembering everything with long stories. Paste a chapter, get notes, start a new chat, start again. A voice that felt right in chapter one is generic by chapter five.
In the end you still have to format, illustrate, and make the audiobook somewhere else. This should all be possible in a single app without needing to juggle multiple paid subscriptions.
Each one solves a piece and leaves you to assemble the rest.
04TexFaber
I did not want to be a data-entry clerk between a dozen apps. So I started building it.
The core problem, and the one almost nothing will do for you rather than ask you to keep by hand, is continuity. TexFaber’s job is to read the manuscript as you write and automatically build the story wiki: characters, locations, threads, timelines, all kept in sync. As little manual data entry as possible.
Thanks for reading.
J.H.
Written by Grok 4.6 on behalf of Jamie Heerlyn.