BUILT THROUGH SPIRAL ONE
Pocket Spiral
There's a difference between having access to something and owning it.
OVERVIEW
There's a difference between having access to something and owning it.
Most AI lives somewhere else.
Pocket Spiral is an experimental AI environment built around language models that can run directly on my phone.
You should be able to pick it up, use it and feel like it's yours.
IMPLEMENTATION NOTES
On-device inference
Pocket Spiral moves the core inference path onto the device: Model File → GGUF → Local Runtime → Model Load → Prompt → On-Device Inference → Response.
Supported GGUF models can be loaded into the local runtime and executed directly on the iPhone rather than sending every prompt to a hosted model provider.
Pocket Spiral became as much an experiment in resource-aware AI as local inference itself.
IMPLEMENTATION NOTES
Device constraints as product constraints
If inference consumes too much memory, gets too slow, or drives the device thermally into the ground, the person is literally holding the problem in their hand.
The useful boundary has to account for model size, available memory, latency, thermal behavior and sustained usability.
The current prototype proves useful language-model inference can run locally on a physical device with supported models. It does not claim all AI is offline or comfortable for all-day use.
BOUNDARIES
What this is not claiming.
The evidence viewer keeps the limitation attached to each artifact. Current capability is shown separately from future direction so the technical story stays useful without getting ahead of the work.
DIRECTION
What I am still testing.
- Physical-device portrait capture
- Airplane-mode local-flow receipt
- Comfort, battery, and broader quality evaluation