FAQ

Questions people ask

Short, straight answers.

Can a memory make my agent trusted, or give it permission?

No. Every answer says grants_authority: false and nothing can change that.

Does it use an AI model?

No. Search, attack screening and contradiction flags are all fixed rules. No model, no network calls, no per-token cost.

Is it a vector database?

No. Put one in front if you need meaning-based search. AETHER-33 is the trustworthy record of what was written, where it came from, whether it changed and whether it may be read.

Why “33”?

It has 33 live checks, each testing something the service must never allow. You can watch them run on the 33 checks page.

Has it been audited?

Not yet. There is no third-party audit, SOC 2 or certification. The 33 checks are our own tests.

Do I need an account?

Not during the free preview. Send the same X-CAIN-Visitor value each time to keep your memories between visits.

What does it cost after the preview?

See the CAIN Studio pricing page. Nothing you saved is deleted because the preview ends.