Confidential ComputingConfidentiality

Confidential Computing: Capability Without Surveillance

How a sealed hardware enclave makes full capability and real privacy possible at the same time — provably.

There’s a sentence almost everyone accepts as truth: if you want a genuinely helpful AI, you have to give up your data. We’d like to show you why that sentence is no longer true.

The old dilemma

An AI that helps you day to day has to perceive your home. It has to hear in order to respond to speech. Often it has to see in order to know whether a room is in use or who is at the door. That perception is demanding — and so the large providers moved it into their data centers. Your voice, your images, your habits therefore travel across the internet to someone else’s servers. There they are exposed to processing by others — and, in principle, open to view.

What confidential computing changes

Confidential computing inverts this model. The idea: data isn’t only encrypted in storage and in transit — it stays encrypted and protected even during processing.

What makes this possible is a sealed execution environment (an “enclave”) that modern processors provide in hardware. Our AI runs inside this enclave. Data is decrypted, used, and re-protected only there. From the outside — even for the operator of the system, even for someone with physical access to the device — the contents remain locked away.

Translated to Verity: your data lives locally, at home — nothing is stored in a corporate cloud. And when the powerful AI needs to compute, it does so in the cloud, but inside such an enclave. So even during processing, what concerns you stays sealed: no one can look in — not Big Tech, not us.

The decisive difference: proof, not a promise

Many providers promise privacy. A promise is something you have to believe.

Confidential computing can do more: attestation. The hardware issues a cryptographic proof that establishes which software is running in which sealed environment. If someone alters the software, the proof no longer matches. This makes it technically verifiable that your system does exactly what we say — and nothing else.

Privacy thus moves from a marketing claim to a property of the architecture. That’s the difference between “trust us” and “you don’t have to trust us.”

Why this changes everything

With that, the old dilemma collapses. Your home can finally be both: deeply capable — it sees, hears, understands, and acts — and deeply private — because none of it can be seen by anyone else, and that can be proven.

This foundation carries everything Verity can do. In the stories to come we’ll show what becomes possible on top of it: hearing, seeing, speaking, remembering, and watching over — and each time, the essential thing stays where it belongs. With you.


Confidential by architecture. Provable, not claimed.