What may the brain tell a model?
Treats every model call as a custody crossing with provider classes, expiring authorization, a sole broker, and an append-only ledger.
I02 Inference Gate
CONFIDE
Controlled Inference and Disclosure Governance
What your brain may tell a machine.
01 READ THE SYSTEM
Spec versus field.
Every module keeps the formal idea separate from Spencer’s lived implementation. The interface can be playful without pretending a draft standard has been satisfied.
Inference Gate
My AI-use journal: which models touched which material, under what custody, and what came back.
Draft means draft.
The source is under active development. This page is an interpretation and field notebook, not the normative specification and not a conformance claim.
02 FIELD TOPICS
What belongs here.
These topics are publishing hooks, not formal spec registers. A single field log may belong here and cross-reference several other modules without being copied.
- 01AI-use journalchannel ready
- 02Provider reviewschannel ready
- 03Local vs cloudchannel ready
- 04Custody decisionschannel ready
- 05Assisted-work reviewchannel ready
03 FIELD LOG
Receipts from real life.
Writing, experiments, intake notes, resources, and journal entries are published once and indexed against every system they touch.
No public field logs yet.
The CONFIDE channel is wired. Reviewed journal entries, experiments, and resources will appear here without duplicating them across connected systems.
04 LINKED SYSTEMS
Follow the signal.
No module explains a life by itself. These are the nearest crossings, dependencies, and useful next views.
↔Peer link · Operates insideBLUEPRINTHow is knowledge held?←From linked system · Governed byDEFERWho may decide?→To linked system · Produces residue forTRACEWhat did the tools leave behind?↔Peer link · Constrains memory withRETAINWhose memory is this?↔Peer link · Shares custody rules withSPEAKWhat crosses to another brain?