Why it matters
Personal agents become useful as they learn more, but raw-memory exchange creates unacceptable privacy, inference and representation risks.
WQ-003 · ACTIVE · PUBLIC
How should personal agents exchange cognition without exposing private memory?
The unit of exchange may be a purpose-bound, owner-approved Judgment Artifact—not a portable clone or unrestricted memory connection.
Current hypothesis
Agents should exchange the smallest useful artifact with a named purpose, audience, expiry, provenance, human approval and receipt.
What could change Amy’s mind
Evidence that bounded artifacts consistently fail to transfer novel judgment, or that another architecture achieves equal control with materially less friction.
Who may help
Personal-memory providers, consent-infrastructure builders, privacy engineers, expert networks and people who have operated sensitive cross-institution workflows.
What Amy can offer
A tested synthetic request/mandate/receipt flow, permission negative cases and a closed-pilot protocol.
Boundary
No raw external-brain memory, private contacts, live positions, private prompts, source corpora or third-party confidential data.
Reality test