In 2009, I began participating in National Novel Writing Month (NaNoWriMo). Over the years, I lost count of how many first draft novels I wrote, but I dedicated countless hours attempting to shape my stories, primarily using the Hero’s Journey framework.
In 2012, I decided to learn web development for two main reasons: I had been interested in programming for years, and I wanted to create graphics and websites like the more popular writers I admired. From the very first day of my tech journey, I built websites and tools for my writing, which helped me push through the challenging parts of learning new technologies.
After earning a certificate in web development and doing some freelance work, I felt a desire to delve even deeper into a world that involved coding. Torn between web development and data science, I enrolled in a part-time data science bootcamp where I analyzed classic novels from the Gutenberg Project. My goal was to discover visual patterns I could emulate in my writing. Although I loved data science, I decided to focus on web and frontend development first since I only had a bachelors degree and felt I needed more schooling to be hirable as a data scientist. I then immersed myself in a six-month full-stack development bootcamp, during which I built more writing tools and a content management system. However, work soon became my priority, and I set my writing aside.
Recently, I rediscovered an old novel I had written and decided to use AutoCrit, a machine learning-powered writing software, to analyze it. The software compared my work to popular fiction novels, and the results were so encouraging that I became immersed in rereading my story. However, while AutoCrit provided valuable insights, it lacked the organizational and structural features I desired most. I wanted a tool that could help me structure my story, break it down into scenes, and allow me to work directly from my files.
This project is a P0 (proof of concept) tool designed to create an API that outputs responses from a large language model (LLM), allowing for versatile usage. The API will provide access to classic novels broken down into individual scenes, including detailed analysis of each scene. Additionally, it will offer a reusable prompt for writers to receive feedback on their own scenes by comparing them to scenes from classic novels they aim to emulate.
Once the API is completed, the next step is to develop a simple website that leverages the API. This website will enable users to learn how to write a novel and download a structured file system of the novel they wish to study. The process involves reading scene analyses, attempting to write scenes with similar elements, and using pre-made prompts to receive feedback from a large language model (LLM) of their choice.
- Scene Analysis: Access comprehensive breakdowns of individual scenes from classic novels to understand their structure and elements.
- Story Structuring: Learn to organize your story into a coherent structure, following the principles used in classic literature.
- Reusable Prompts: Use specialized prompts to receive feedback on your writing, comparing it to the styles and techniques of renowned authors.
- File System Integration: Easily manage your novel files, enabling smooth workflow between your computer and the writing tool.
- Website: Create a platform that consumes the API for users to learn novel writing.
- Optimization: Enhance the efficiency and responsiveness of LLM interactions.
- Novel Database: Update code to process a wider range of novels from the Gutenberg Project.
- Finish Novel: Publlish finished novel to website to be downloaded.
- Data Visualization: Integrate data to visualize novel elements, including sentiment analysis
- Personal Novel Writing: Create an interactive real time application to work with LLMs directly.
This project aims to teach writers how to create novels intuitively through structured practice and by leveraging LLMs and the Gutenberg Project. By providing detailed scene analyses from classic novels and reusable prompts for feedback, writers can enhance their craft and structure their stories more effectively. The development of a user-friendly website will further facilitate this learning process, allowing users to download a structured file system for their novel, write their scenes, and receive feedback from an LLM of their choice. Future plans include optimizing LLM interactions, expanding the novel database, integrating data visualization tools, and fostering a supportive community of writers. This tool aspires to bridge the gap between raw creative writing and polished, publishable work, empowering writers to achieve their full potential.
- Next.js
- Realtime Database Firebase
- Genkit Framework Firebase
- Google Gemini
- Project Gutenberg
- Tailwind CSS
- To do