Personal intelligence
Also known as: personal context layer, personal data access, cross-app personal context
Personal intelligence is the term Google uses for the permission layer in Gemini Spark that lets the agent access your Gmail, Calendar, Drive, Docs, YouTube history, and other Google services. But the concept is broader than one product. Any always-on agent that needs to act on your behalf in a meaningful way needs some version of this: a structured way to connect to the personal data that gives context to what 'help me plan my week' actually means for you specifically.
The design challenge here is trust. The same data access that makes an agent genuinely useful, knowing your recurring commitments, understanding your communication style, noticing that a meeting conflicts with a flight, is also the data access that creates real privacy exposure if something goes wrong. Well-designed personal intelligence layers are opt-in per service, transparent about what's being read, and revocable at any time. Google surfaces this as a settings menu; other implementations use OAuth flows (a standard protocol that lets you grant an app limited access to another service without giving it your password).
For builders designing agents that will touch user data, personal intelligence is the architectural decision underneath everything else: what data does the agent see, how granular is the permission model, and what happens when the user revokes access mid-task? Those questions are landing in product specs and privacy reviews today in a way they weren't eighteen months ago.