I’m the maker of Book Factory. I stopped building another cloud AI writer because I was tired of renting the tools I use to think and write. Subscription fatigue is real when a writing project can take months and the work is personal. I wanted ownership instead: a production line that runs on the author’s machine, keeps the project in Documents/Book Factory as plain-text files, and doesn’t turn a book into another monthly bill.
Book Factory is a local-first novel production system for Mac, Windows, Linux, and Android tablets. You open it in a browser on localhost, while the actual work stays on your machine. Your words and files remain where you can manage, back up, and inspect them—not trapped in a hosted editor. It works for fiction and nonfiction, and you bring your own OpenRouter account. A full novel costs about $3 in model usage, depending on the models and how much you revise. The app is a one-time purchase with no subscription markup.
The workflow is deliberately structured so the author stays in charge:
That approval step is the part I care about most. The system can help with the labor, but it should not quietly decide what your book is. You get a chance to shape the structure before the words arrive, then give notes between passes.
The goal is not “press a button, receive a novel.” It is a repeatable process that makes a large creative project easier to direct and finish while leaving the author’s judgment in the loop.
I built this because I wanted a practical alternative to sending every draft to a hosted writing service and hoping the subscription still makes sense next month. Local-first also means you can choose the model provider you trust and keep working with ordinary files. I’m not claiming magic or inventing metrics; this is simply the tool I wanted for owning the work instead of renting access to it.
If you’re curious, I made a free checklist covering the local novel-production setup and the decisions behind the pipeline: https://magmafire60.gumroad.com/l/gdbvmp
For people who want to try Book Factory itself, there is a limited beta at $174.50—50% off the $349 founding price: https://magmafire60.gumroad.com/l/tdpmtw
I’d especially value feedback from authors who want more control over their process, whether you’re building a fictional series or working on nonfiction. What would you want to own locally instead of rent in the cloud?
Curious how long it took before you saw the first real results?
Makes sense. Are you planning to charge for it, or keep it free for now?
How did you decide this was worth building in the first place?
Good point. Did you test that with users before committing to it?
This is useful. How are you finding your first users so far?
Appreciate the honesty here, most people only share the wins.
The approval gate is the real product insight here. Long form generation needs checkpoints around structure and canon, otherwise a fast model just makes expensive wrong turns at scale. Keeping the files local also makes it easier to swap models without turning the workflow into a subscription dependency.
Clear and practical, thanks. Did anything surprise you along the way?
Clear and practical, thanks. Did anything surprise you along the way?
Clear and practical, thanks. Did anything surprise you along the way?
The biggest surprise was how much the approval gate mattered once real manuscripts were in play. Faster generation was never the hard part — keeping the author’s plan as the source of truth was. Local plain-text files also made debugging and backups boring in a good way.
Interesting take. Would you still recommend this approach to someone starting today?
Yes, if you want ownership of the process and files. If you mainly want chat-style drafting with zero setup, a cloud tool is still fine. I’d start with the free checklist and only buy if the gated pipeline matches how you already work: https://magmafire60.gumroad.com/l/gdbvmp
Interesting. How are you measuring whether it is working?
Right now: founding beta users (aiming first 20–30), where people stall in the funnel (setup → bible approval → first export), and whether the free checklist helps as a lighter first step: https://magmafire60.gumroad.com/l/gdbvmp. Not inventing vanity metrics — early feedback is the signal.
What made you pick this stack over the alternatives?
I wanted local files I can back up, BYO models via OpenRouter (no markup), and a one-time purchase instead of another subscription. The alternatives I kept bouncing off were either chat boxes with weak continuity or hosted tools that own the project.
Nice work shipping it. What has been the biggest challenge since launch?
Getting the first real users and learning where they pause — marketing/distribution more than the core pipeline this week. Product-wise, keeping “plan before pages” obvious without making the flow feel heavy.
Nice progress. What is the next thing you are focusing on?
Near-term, I’m focused on getting the first 20–30 founding beta users and learning where they stall. I’m using the free checklist (https://magmafire60.gumroad.com/l/gdbvmp) as a lighter first step, then inviting the right people into the limited beta (https://magmafire60.gumroad.com/l/tdpmtw).
Your approval gate is a strong differentiator from “press button, get novel.” I’d instrument setup → story-bible approval → first export and ask where people pause; that will tell you what to simplify before adding more passes. I built SoftShelf; my free 3-debt snowball mini is here: https://softshelfstudio.gumroad.com/l/fgyrkj?utm_source=indiehackers&utm_medium=community&utm_campaign=first_sale&utm_content=comment_free. If comparing Snowball vs Avalanche with a dashboard helps, the full planner is $9.99: https://softshelfstudio.gumroad.com/l/eqdyqi?utm_source=indiehackers&utm_medium=community&utm_campaign=first_sale&utm_content=comment_paid
Thanks — the funnel instrumentation idea is really useful. I’ll watch where people pause between setup and story-bible approval, and again before the first export; early beta feedback should be the next data source for a light pass on the flow.
Clear and practical, thanks. Did anything surprise you along the way?
The biggest surprise was how much the approval gate mattered once real manuscripts were in play. Faster generation was never the hard part — keeping the author’s plan as the source of truth was. Local plain-text files also made debugging and backups boring in a good way.
What made you pick this stack over the alternatives?
I wanted local files I can back up, BYO models via OpenRouter (no markup), and a one-time purchase instead of another subscription. The alternatives I kept bouncing off were either chat boxes with weak continuity or hosted tools that own the project.
Really relatable. How much time do you put into this each week?
Varies. Building/shipping was concentrated; now more of the week is support, outreach, and listening to where people stall. Still a side project alongside a full-time job, so I protect evenings/weekends for the highest-leverage next step.
Good point. Did you test that with users before committing to it?
I dogfooded the pipeline myself and used early testers / checklist feedback more than a formal pre-commit study. The limited beta is specifically to learn stalls before scaling — that’s the real user test.