solstone support

three ways to process your journal

solstone 2026-11-21 17:02:10

three ways to process your journal

you choose how your journal gets processed, during setup, and you can change your mind anytime in the journal's thinking app.

local, the default

on a capable computer, processing happens on your own device: a bundled local model. capable means enough GPU memory (on linux or windows; on linux that's amd, nvidia, or intel) or enough memory on apple silicon. setup checks your computer against that bar. see what your machine needs to think locally for the full detail.

nothing leaves your device on this lane. transcription runs on your device too.

your own model

bring your own key from Claude, Gemini, or GPT, use your own ChatGPT plan, or point the journal's thinking app at any OpenAI-compatible endpoint you run yourself. your billing, your control. a key stays in your journal, and sol pbc is never in the path. with your own model, your computer talks to the provider or endpoint you chose directly, and it sees the parts of your journal that processing uses.

confidential processing, a paid service

processing on confidential hardware sol pbc runs, off your own device, without a third-party AI provider in the path: no content retained, no human review, nothing used to train. your thinking leaves your device, and your journal itself never does.

the hardware behind this is confidential GPU hardware sol pbc operates on Azure. see confidential processing on the services portal for the full disclosure.

turn it on from the journal's thinking app. with journal 2.0.38 or later, if it isn't on yet, the thinking app takes you to the services portal to finish turning it on. the services portal shows what it costs. it's complimentary for approved scouts.

switching

on the computer your journal runs on, open your journal and go to its thinking app to switch. one lane is active at a time.