Work · Private product
StoryArbor
A native macOS app for private story intelligence. It ingests a manuscript locally, detects chapters, answers questions with citations, and analyzes continuity, with inference running on the Mac.
- Role
- Creator and engineer
- Status
- In development, not released
- Stack
- SwiftUI, Rust core, FFI, Local inference
- Links
Chapter 01
What it is
StoryArbor is a Mac app for working with a long manuscript. It ingests the manuscript on your machine, detects the chapters, and lets you ask questions about your own story. Answers cite the passages they came from. It also runs continuity analysis across the text.
How it’s built
- A SwiftUI app. The interface is native macOS.
- A Rust core. The core is a Rust library that the app calls through a foreign function interface.
- Local inference. Models run on the Mac.
The decisions that shape it
Source folders are read-only. The app treats the manuscript’s source folder as read-only, so analysis cannot alter the original files.
Answers carry citations. A question about the story comes back with the passages that support the answer, so it can be checked against the text.
Local by default. Ingestion and inference run on the machine.
Status
In development and not released. The source is private. The product site is a coming-soon page.
Chapter 02
Local by default. The manuscript, the models, and the answers stay on the Mac.
One machine
- The programAn app such as StoryArbor, and the code inside it that loads and runs the model.
- The graphics backendMetal on a Mac, CUDA on a machine with an NVIDIA card. This is where the arithmetic runs.
- The model weightsA large file of learned numbers, loaded from the local disk.
Prompts, documents, and outputs stay on that machine.
Chapter 03
Sources
Public references for the claims on this page.