Inspect
Read what Guardian holds about you in plain language — not a raw log, an understandable account.
Privacy-first AI
Privacy-First AI That Keeps You in Control
AI assists. You decide.
Privacy-first AI means the assistant's memory belongs to you: you can see what it knows, correct it, limit how it is used, export it, and delete it. Guardian Daily is built that way — memory by permission, authority granted in narrow scopes, and no capability that widens on its own as the system learns.
Opens the Guardian Daily app. You may need to sign in or create an account.
The principle
Knowing and doing are separate powers
An assistant useful enough to help with your life necessarily learns about your life. Guardian's answer is not to know less than it needs, but to keep knowledge and authority apart and to leave both under your hand.
Control
What you can actually do
Correct
Fix what is wrong. Your correction overrides what Guardian inferred.
Limit
Restrict what may be used for a purpose, a person, or a situation.
Export and delete
Take your information with you, or remove what should not persist.
Permissions
Scoped, reviewable, revocable
Permission is not a single switch. It is specific to what, for whom, and within what limits — and it can be withdrawn without argument.
- Authority is granted for defined situations rather than granted broadly and trusted to be used well.
- Every permission can be reviewed later in the same terms it was granted.
- Revoking access takes effect on future actions, and Guardian says so plainly.
- Sensitive information gets restraint by default, even when access exists.
Restraint
No silent authority expansion
Systems drift when capability grows quietly. Guardian is built so that improvement never becomes permission.
Learning stays knowledge
Better understanding of your patterns does not entitle Guardian to act more widely.
Uncertainty is disclosed
When Guardian is unsure, it says so instead of presenting a guess as fact.
Actions are auditable
What was done, under whose authority, and what actually resulted.
Questions
Common questions
- What does privacy-first AI mean here?
- It means memory exists by permission rather than by default, that you can read what the assistant knows in plain terms, and that no capability quietly widens over time. Privacy is a property of the design, not a setting added afterwards.
- Can I see and correct what Guardian knows about me?
- Yes. Guardian's memory is inspectable in readable language. If something is wrong, you correct it and your correction is treated as authoritative.
- Can I delete or export my data?
- Yes. You can export what belongs to you and delete what should not persist. Deletion is meant to be real, not hidden behind a label.
- Does Guardian gain more authority as it learns?
- No. Learning improves understanding; it never expands permission. Wider authority requires a new, explicit grant from a person.
Keep reading
Explore Guardian by need
Four short guides covering the questions people ask most before they start.
Human-in-the-loop AI
How approvals keep consequential decisions with people.
Personal AI assistant
Daily planning and open loops, with memory you control.
Family AI assistant
Household coordination with boundaries between private and shared.
Trust and governance
The commitments Guardian holds itself to, written down.
Your information. Your limits. Your call.
Guardian is designed so that control is the default, not a request you have to make.