Mind the GeekLunch Talk40min
From Developer to Story Builder: Engineering a Novel, an ARG, and a Transmedia World with AI
A technical case study on using software-engineering practices to co-author a novel with AI. It covers context engineering, spec-driven workflows, role separation, governance to prevent drift, automated publishing, and an ARG extension, offering practical methods for managing long-running AI creative projects while preserving human authorship.
talk.summaryAiDisclaimer
Raphael SemeteysWorldline
talkDetail.whenAndWhere
Tuesday, October 6, 12:40-13:20
TBA 2
talks.roomOccupancytalks.noOccupancyInfo
What happens when a software architect treats a novel like a production system?
In this talk, I will share how I built The Human Protocol, a cyber-realist techno-thriller about AI, using the same engineering principles I apply to software architecture: context engineering, spec-driven development, harness design, versioning, review loops, and automated publishing pipelines.
The session is not about “asking ChatGPT to write a book.” It is a technical case study in designing a controlled human-AI creative workflow. I will show how I created a lore-as-code repository, a governance package to prevent context drift, specialized AI roles for design, drafting and review, and a DevOps-style pipeline to produce manuscript formats for publication.
I will also cover the project’s transmedia extension: an ARG website where clues hidden in the book unlock in-universe content. This was built through spec-driven development anchored in the same lore repository, turning the manuscript into both a story and an interactive system.
You will leave with practical patterns for managing long-running AI-assisted projects: how to structure context, debug hallucinations and inconsistencies, separate agent roles, preserve human intent, and ship complex creative artifacts with AI without losing authorship.
In this talk, I will share how I built The Human Protocol, a cyber-realist techno-thriller about AI, using the same engineering principles I apply to software architecture: context engineering, spec-driven development, harness design, versioning, review loops, and automated publishing pipelines.
The session is not about “asking ChatGPT to write a book.” It is a technical case study in designing a controlled human-AI creative workflow. I will show how I created a lore-as-code repository, a governance package to prevent context drift, specialized AI roles for design, drafting and review, and a DevOps-style pipeline to produce manuscript formats for publication.
I will also cover the project’s transmedia extension: an ARG website where clues hidden in the book unlock in-universe content. This was built through spec-driven development anchored in the same lore repository, turning the manuscript into both a story and an interactive system.
You will leave with practical patterns for managing long-running AI-assisted projects: how to structure context, debug hallucinations and inconsistencies, separate agent roles, preserve human intent, and ship complex creative artifacts with AI without losing authorship.
Raphael Semeteys
I am a technologist and Content Producer with over 25 years of experience across software engineering, open-source communities, and emerging tech. My current focus is on the frontiers of Generative AI, specifically local LLMs, image generation (Stable Diffusion), and Agentic AI.
Throughout my career, I’ve balanced technical leadership with a passion for knowledge sharing. I created the QSOS method for open source evaluation and recently explored the intersection of "vibe-coding" and storytelling by publishing AI-assisted techno-thrillers.
A frequent speaker at international conferences like Devoxx, Open Source Experience, and various DevFests, I enjoy deconstructing complex ecosystems—from Web3 architecture to multimodal AI—into clear, actionable insights. I also bring a unique perspective to tech by integrating holistic yoga practices into my approach to innovation and engineering.
More about me: https://raphiki.github.io/
Throughout my career, I’ve balanced technical leadership with a passion for knowledge sharing. I created the QSOS method for open source evaluation and recently explored the intersection of "vibe-coding" and storytelling by publishing AI-assisted techno-thrillers.
A frequent speaker at international conferences like Devoxx, Open Source Experience, and various DevFests, I enjoy deconstructing complex ecosystems—from Web3 architecture to multimodal AI—into clear, actionable insights. I also bring a unique perspective to tech by integrating holistic yoga practices into my approach to innovation and engineering.
More about me: https://raphiki.github.io/