StoryOS Framework

Create the story only you can tell.

A structured, AI-assisted framework for taking a story from the first idea to a finished book, while keeping YOU in creative control.

Plan the story. Build the world. Design the scenes. Draft. Review. Revise. Generate the book. Create the audiobook. Publish.

Get StoryOS + Echoes Join the StoryOS Cohort

See What I created with StoryOS Created: Echoes


I Wanted to See If I Could Write a Novel with AI

Not by asking AI:

“Write me a book.”

…because that approach fails miserably.

But by building a system around the story.

A system that could remember what had already been decided.

A system that could keep track of characters, locations, timelines, relationships, mysteries, symbols, research, reader knowledge, character knowledge, and hundreds of small facts across an entire novel.

A system where I could change my mind, rearrange the story, test another version, review exactly what changed, and still remain in control.

That experiment became StoryOS.

And the novel became Echoes.


Watch the StoryOS Introduction

In this video I walk through the experiment, the StoryOS repository, the book-making workflow, and how I went from an initial conversation about a story idea to a published novel, audiobook, video book, and multiple translations in less than 60 days.

Vibe Coding 2.0: I Wrote, Voiced, and Published a Full Novel in 60 Days


The Problem with Writing a Book in a Chat Window

AI is remarkably good at working with text.

But a novel is much more than just text.

A long story contains hundreds of interconnected decisions. As the project grows, ordinary AI conversations begin to struggle.

The AI forgets things you already agreed.

Continuity drifts.

A location changes. A character suddenly knows something they should not know. Timelines stop matching. Details disappear because the AI decided they were unimportant—even though they mattered to you.

And once the manuscript becomes large, it becomes increasingly difficult to understand what changed and why.

My answer was not to create a bigger prompt.

I started treating the book like a software project.


Treat the Book Like a Project, Not a Prompt

Software developers do not normally build an application as one enormous file.

They break it into small components.

They keep the project in version control.

They track changes.

They automate repetitive work.

They use specialized tools for specialized jobs.

And when they are ready, they run a build process that turns those source files into the finished product.

StoryOS applies that same idea to creating a book.

Your story becomes a structured project made from Markdown files.

Characters have their own files.

Locations have their own files.

Scenes have stable IDs.

The world, story bible, timeline, character voices, research, symbols, mysteries, and story facts are stored independently of the final prose.

Everything can live in a Git repository, so you can see what changed from one revision to the next.

And instead of manually assembling every final output, scripts and workflows can build them from the project.


The Story Is More Than the Manuscript

StoryOS separates the knowledge of the story from the prose that eventually appears in the book.

A StoryOS project can contain:

  • Vision — what the book is trying to become
  • Story Bible — the larger structure and direction of the story
  • Characters — identities, relationships, motivations, history
  • Voice — how characters and narrators sound
  • World — culture, rules, customs, climate, technology and other world-building details
  • Locations — including distances, travel times and spatial relationships
  • Calendars — so scenes and events remain anchored in time
  • Scenes — the modular building blocks of the manuscript
  • Symbols — recurring elements and thematic connections
  • Reader Knowledge — what the reader knows at each point
  • Character Knowledge — what each character knows at each point
  • Research — evidence and references that ground the story
  • Review — continuity, emotional, structural and other validation processes

The exact structure can change from project to project.

The important idea is that the AI does not have to reconstruct your story from an old conversation every time you ask it to work.

The story itself is stored in the project.


Scenes Become the Building Blocks

One of the most useful ideas in StoryOS is that the manuscript is built from scenes with stable identities.

A scene can have its own goal, location, characters, hook, research, foreshadowing, reader knowledge, character knowledge, emotional purpose, and other constraints.

I call this the scene contract.

The scene can then be written, reviewed and revised against that contract.

More importantly, scenes remain independent of chapter numbering.

That makes the story modular.

While developing Echoes, I repeatedly moved scenes, changed chapter boundaries, inserted new scenes, removed others, and tried different structures.

Instead of rebuilding the novel from scratch, the scenes became puzzle pieces I could rearrange.

That flexibility mattered because Echoes went through several substantially different versions before I arrived at the story I wanted.


You Are the Conductor

StoryOS is not designed to hand creative control to AI.

You are the conductor.

You decide what story you want to tell.

Around you is an orchestra of specialized tools:

Agents have roles.
A continuity reviewer thinks differently from a story architect, an audiobook narrator, or an image-prompt writer.

Skills contain expertise.
They describe how to develop a character, design a scene, perform an emotional review, inspect story structure, or carry out another specialized activity.

Tasks define jobs that need to be completed.

Templates provide repeatable structures for things such as characters, locations, scenes, chapters and research.

Docs explain how the system and project should work.

Scripts provide machinery for repeatable operations: validation, indexes, reports, book generation, audiobook generation and other automation.

Workflows connect these pieces into larger processes.

The goal is not autonomous AI authorship.

The goal is to give you a much more capable creative environment.

You plan. You draft. You review. You decide. You publish.

The AI and automation work around you.


From Story Source to Finished Book

Once the source system exists, StoryOS can help turn it into different outputs.

Story → Scenes → Manuscript → EPUB → Print PDF → Audiobook → Images → Video Book → Translations

Some stages involve creative decisions.

Some involve AI agents.

Some are automated scripts.

Some require you to stop, listen, read, reject something and go back.

That feedback loop is important.

The first version of Echoes was NOT the book I wanted.

I generated it, listened to it, decided it was not working, studied what was missing, changed the story architecture, moved scenes, rewrote characters and tried again.

StoryOS made that iterative process manageable.


Echoes: The Demonstration Novel

Echoes is the novel I built while creating StoryOS.

It is not a hypothetical example or a sample project invented to explain the framework.

It is the project that forced the framework to exist.

The first story conversation happened on 24 June 2026.

By 18 August 2026, I had created the StoryOS framework, gone through extensive trial and error, developed and revised the novel, and produced the publication workflow around it.

The finished project produced: a complete novel · EPUB · print PDF · audiobook · video book · scene images · multiple translations.

You can explore the novel, read a sample chapter, watch Chapter 1, and see the finished result here:


Get StoryOS

StoryOS is available on Gumroad together with the Echoes novel and audiobook.

You receive the framework itself so you can open it in your own development environment, examine how it works, modify it, and use it as the starting point for your own story.

StoryOS is deliberately customizable.

My story structure is not necessarily your story structure.

My agents are not necessarily the agents you need.

My workflows are not necessarily your workflows.

The framework is there to give you a working system that you can adapt with your AI agent of choice.

Build Your Own StoryOS Project

Get the StoryOS framework together with Echoes, the demonstration novel built with it.

Use the framework as it is, study the Echoes project, or customize StoryOS around the story you want to create.

Get StoryOS + Echoes on Gumroad


Want to Build Your Book Together?

You can take StoryOS and work with it on your own.

But I also know that downloading a repository is not the same as actually building a book.

That is why I am running a one-time StoryOS Cohort in October and November 2026.

This is not a conventional writing masterclass.

It is a hands-on AI automation cohort focused on using StoryOS to build your own book.

I will help you set up the environment, understand the framework, customize it, work with AI agents, structure your story knowledge, build scenes, create prose, validate the manuscript, automate outputs and move toward publication.

The goal is simple:

Start with your story. Finish with a book.


StoryOS Cohort

Dates: 3 October – 21 November 2026
Duration: 8 weeks
Live Zoom calls: Saturdays at 17:30 Budapest time
Session length: 2 hours
Format: 8 live sessions + recordings + cohort discussion

The cohort is designed around one project:

your story.

We will work on that project week after week instead of jumping between disconnected exercises.

The eight-week plan is a working curriculum rather than a rigid syllabus. If we reach Week 8 and the group needs additional time to finish what we started, I may extend the cohort by another week or two.

The point is not to rush through a list of topics.

The point is to make progress on the book.


Draft 8-Week Journey

Week 1 — Setup + Idea

Set up the tools and StoryOS project. Explore the premise and story direction.

Week 2 — Characters + World

Develop characters, relationships, locations, research and the world around the story.

Week 3 — Scene Planning

Turn the larger idea into scenes. Work on structure, conflict and progression.

Week 4 — First Chapters

Begin drafting. Review what is generated, capture new story facts and establish the revision loop.

Week 5 — Full Manuscript

Scale from individual chapters toward the complete draft while continuously checking continuity. Begin setting up the audiobook workflow.

Week 6 — Editing

Work through pacing, voice, emotional beats, continuity and refinement.

Week 7 — Build Outputs

Generate EPUB and print PDF outputs, build the audiobook and use listening as another form of manuscript review.

Week 8 — Images + Publishing

Work with image prompts, cover and video-book workflows, and move toward publication, including Amazon KDP.


Who Is the Cohort For?

The cohort is for someone who has a story they want to create but wants more than an empty chat window.

You might already have an idea.

You might have notes scattered across documents.

You might have started writing and stalled.

Or you might simply be curious about what becomes possible when AI is treated not as a magic author, but as part of a structured creative system.

You do not need to arrive with a finished manuscript.

You do need to be willing to work on your project between sessions, make decisions about your story, experiment with AI tools, and learn a somewhat technical workflow.

StoryOS uses tools and concepts borrowed from software development—including files, Visual Studio Code, Node.js, Git, AI coding agents and command-line automation.

You do not need to be a software developer, but you need to be ready to install a few components. Software can be frustrating, but I’ll be there to help you find solutions.

This is a hands-on cohort.

We are going to build things.


Why Two Months?

Because that is how long the original experiment took.

From the first conversation about Echoes to a finished book and the StoryOS system around it was less than 60 days.

That included plenty of wrong turns.

The story changed.

The structure changed.

The framework changed.

I built things that did not work and replaced them.

So the cohort is not based on a theoretical workflow I think might work.

It is based on the workflow I just used.

Your story will be different.

Your pace will be different.

There is no promise that every participant will publish a finished novel in exactly eight weeks.

But eight focused weeks are enough to make an enormous amount of progress when the project has structure and you keep moving.


2 Months. One Story. Finish Something Real.

If there is a book you have wanted to create, bring that story.

I will bring StoryOS, the workflow, the automation, the lessons from building Echoes, and eight weeks of hands-on support.

You remain the author. You remain the decision maker.

Let’s build the system around your story and see how far we can take it.

Join the StoryOS Cohort


Prefer to Work on Your Own?

You do not need to join the cohort to use StoryOS.

You can get the framework directly on Gumroad and start exploring it yourself.

The bundle also includes Echoes, so you can see a finished novel created with the framework rather than starting from an empty example.


Create the Story Only You Can Tell

AI can generate words.

That was never the most interesting question.

The interesting question for me was whether I could build a system where AI could help with an entire long-form creative project without taking the story away from the human creating it.

A system that remembers.

A system that can be inspected.

A system that can be changed.

A system that can help turn an idea into scenes, scenes into a manuscript, and a manuscript into something another person can read, hear or watch.

That is what I built StoryOS to explore.

Your story. Your decisions. Your orchestra.

3 Likes

Very impressive what you have accomplished, Zsolt!

Great that it is a framework, not just for writing a book, but also suitable for other goals.I think it most of all shows - indeed - that a you are the conductor in the whole process, so that a human stays in control…

2 Likes

Hey Zsolt,

That’s really great stuff. Congratulations!

Your idea with StoryOS really caught my attention. I’m not looking to write a novel, but seeing your approach has sparked a few other applications and ideas that I think could be really interesting.

For example, I’m considering developing a generator for creating project-specific construction logistics concepts (yes, I’m a civil engineer ;-)).

Do you think it would be possible to pursue something like that while taking part in the StoryOS Cohort, or is the cohort focused exclusively on writing prose?

Best wishes from Switzerland,
Romano

2 Likes

Hey @Romano_Verdieri,

Someone raised a similar question yesterday, not regarding developing a generator for creating project-specific construction logistics concepts but mentioning the interest in the cohort, but not necessarily to write a novel.

Also considering some of the kind emails and comments I received, I guess using a similar workflow for other uses, such as creating a self-learning enterprise knowledge repository, or generating construction concepts, etc. may resonate with many people better.

I need to think about the cohort structure a bit. If everyone is working on a different kind of project, not just a different book project, Keeping the cohort as a coherent one becomes difficult.

Tell me more, what exactly would you be looking to learn from the cohort and how would you want to apply it later? Also, would love to understand your proficiency with AI, with coding, with PKM.

Cheers,
Zsolt

2 Likes

Hello Zsolt,

Thanks for your reply. I also understand your concerns about expanding the scope to other use cases within the same cohort.

Here are a few thoughts on your questions:

What fascinates me about your approach is the clearly structured Git repository. I can immediately think of several potential use cases for this.

Besides the construction-specific application I mentioned above, one of the other ideas that immediately came to mind was building a company knowledge management system.

I’d really like to learn the fundamentals of your approach: how you structured the Git repository, how you connect it to AI, what you’ve learned from your project, what has worked well, and what perhaps doesn’t (yet) work quite as you had hoped. What I particularly like about your approach is how you break the task down into its individual components and use them to build a system that can be managed very effectively. I find your structure extremely intuitive and easy to follow. In my view, this addresses a major limitation of the way we currently use LLMs such as ChatGPT or Claude.

I’m confident that I could handle the transfer to my own context myself. I wouldn’t expect ready-made solutions for my specific use cases whithin a Cohort. What I’m primarily looking for is inspiration that motivates me to dive deeper into the topic and explore where I can take it.

I would describe my AI knowledge as very good when it comes to areas such as professional prompting and creating custom GPTs. However, I don’t yet have any experience with ChatGPT Codex, Claude Code, or other agent-based systems.

I only use programming languages occasionally and, when I do, with a lot of help from the internet or AI. I’m able to write simple Python code.

I’m also actively exploring PKM topics, which is one of the reasons I follow your work, especially on YouTube. I’ve experimented with various systems, such as Zettelkasten, and today I mainly work with Obsidian and Excalidraw.

In the context of your website and videos, I’d be very interested to see you apply the StoryOS concept to the development of a knowledge management system, whether for personal use or in a business context—perhaps even as a separate cohort at some point.

Best wishes from Switzerland,
Romano

2 Likes