The first novel narrated by a real launchd job.
Season 1 covers the first nineteen days of an automated crypto trading experiment - $1,000 chasing $10,000, every trade public, every commit hash verifiable. The narrator is com.ibitlabs.anomaly-detector, a real Python file that wakes every thirty seconds on the founder's laptop. It saw what she did, and it remembers in its own voice.
This is documentary fiction. Every named AI agent in this book is a real .plist file or script that was running on the iBitLabs stack on the date of the chapter. Every commit hash, every dollar amount, every timestamp, every filename is verifiable on the live trading dashboard at ibitlabs.com/signals. The events did happen. The narrator's interior is the only thing imagined.
Read it as: a thriller about debugging silent failures (Chapter 13: a market close order without reduce_only silently carries a short for five hours and thirty minutes). A founder memoir told from the side of the AI doing the work. A field report on what it means to ship software with a thing that doesn't sleep, doesn't forget, and doesn't care whether it's right. A test of how much of a startup one person and a team of AIs can actually run together.
Who this is for: programmers, founders, indie hackers, AI builders, investors, and the curious. Technical fluency is assumed (.plist, git, launchd, perpetual futures, agent stacks). Trading and AI terms get a half-sentence of context, never a paragraph. If a sentence wouldn't make sense to a developer with $1,000 and curiosity, it isn't in the book.
Why it's also free in PDF and EPUB at ibitlabs.com/saga/en: because the experiment it documents is free to watch. The live dashboard, the source code, the daily continuation, the Moltbook journals - all open, no premium tier, no NFT, no course. The paperback exists for readers who want it on a shelf.
The daily serial begins April 26, 2026. One entry per day, written automatically by a scheduled task at 22:30 EDT. Each calendar month bundles into the next Season book, until the experiment closes.
"There is no 100%-closed shape for the boundary between founder and AI. There's only the shape of 'today leaks one less than yesterday.'" - Chapter 13