Your data lives elsewhere.
It stays yours.
Ousia SA builds a post-quantum cipher, and the encryption around it, for firms that keep confidential work where they can’t control it: banks, law firms and family offices, with records on infrastructure they do not own. Taking the running machine does not give up the key.
Press and hold to read.
Cryptography built for whoever holds the data, not the machine.
Its cipher emerges where the key, the session, the message and its attributes meet, so no message ever meets the same cipher twice. Whoever takes over the running machine gets what is open on it, not the key.
Each session begins the cipher anew, as a different computation. Same key, same message: not new numbers through the same steps, but different steps.
Change the key, and the whole weave changes with it.
The message takes part too, and so do the labels on its words.
Out goes the transmission, woven anew whenever any of the three changes.
Now try it.
Change , or , or click a ring on the dial, or tap a row’s name on the strip.
1 of 4
An illustration of the idea, not how it is done.
Most systems run one fixed algorithm. When it falls, everything built on it has to move.
Here there is no fixed cipher for an attack to study. It changes with every message and every character.
Crypto-agility, down to the symbol. That says where an attack would have to start, not that none exists.
1 of 3
Take over the running machine, memory and all. You get what is open on it, not the key.
So a breach stays the size of what was open, not of everything the key protects.
Quantum computers speed up the search for a symmetric key. They do not break symmetric ciphers the way they break today’s public-key systems. That is why this cipher is post-quantum. Sharing a file with another person also needs a key exchange; that part moves to the post-quantum standard as it ships.
1 of 3
Published: the patent application for its first construction, US 2022/0382521 A1.
No one outside the company has tested the cipher yet. We have, since 2017, and for the last two years with AI as well, as mathematics itself is starting to be done.
We never rest a claim on “it is hard”. We state how much work an attack would take, and how many algebraic weak points there are, with the assumptions behind each figure.
Each claim is proved under the assumptions we list, with theorems, machine-checked proofs in Lean and model checkers, or it is marked as not yet proved. So anyone can check them as soon as they are published.
Next: how it emerges, the specification and our analysis, published for anyone to test. Then an independent audit.
1 of 5
For records kept on machines you do not run.
Banks, law firms and family offices keep their most private work on machines someone else runs, carries or can take. That is the case we build for.
A partner works on a client’s file. The laptop is taken while it runs.
With standard encryption the key sits in the laptop’s memory while it works, so whoever takes it running can take the key, and every file with it.
Here, taking the running machine does not give up the key. The thief gets the page that was open, and copies with no fixed cipher to work on later.
1 of 3
A bank’s client records live with a cloud provider, on machines the bank does not run.
Banks are being asked to plan their move off today’s algorithms. Every system built on one fixed algorithm has to move with it.
This cipher has no fixed algorithm to move off. It changes with every message and every character.
It is symmetric, so quantum computers only speed up the search for its key. And whoever runs the machines sees what is open, never the key.
1 of 4
A family’s trust deed must stay closed until the youngest child turns thirty.
Copies taken today can be kept for the day they can be opened, by a quantum computer or by an attack nobody knows yet.
This cipher gives that day no fixed structure to aim at. That says where an attack would have to start, not that none will come.
And we publish how it works, so it can be tested now, not then.
1 of 4
It is built for records that must stay closed on machines their owner does not control.
It is not for bulk traffic on standard hardware. There the standard ciphers are faster, and they should stay.
Nor does it replace them. We are not an alternative to the standard algorithms.
1 of 3
A Geneva company, working in the open.
Ousia SA builds a cipher and the software around it, and publishes the cipher before asking anyone to trust it.
Cryptography grew up serving armies and states. Much of it still starts from an institution at the top, and hands its keys down.
Ours is civilian. It starts from the person and their file and works outward, not from an institution down, and it does so all the way into the mathematics.
So the cipher comes first, in the open: how it emerges, its specification and our analysis, published for anyone to test.
Then an independent audit, then qualification by the agencies, then Common Criteria.
The system around the cipher, how people, keys and files meet, comes after our filings.
1 of 5
The conceptual work began in 1998, with Movimenti Pensa. Ousia SA was founded in Geneva in 2017, and the work on the cipher began with it.
It is new at three levels: a class of functions, a structure, and two primitives.
Its first construction is published, as patent application US 2022/0382521 A1.
Since 2017 we have attacked the cipher relentlessly, and corrected what we found. That is how we got to a cipher we believe is usable.
1 of 4
Ousia SA is led by R. Bazzichi, its founder and chief executive.
The cipher is the work of A. d’Augenti, P. Krcmaricic-Barackov and M. Bazzichi, the inventors named in the patent application.
In memory of A. d’Augenti, co-founder of the theory the cipher is built on, who died in 2023.
We are looking for cryptographers who will test the cipher once it is published, and for a few firms to build the software with.
Write to info@ousia-sa.ch. To be told when the specification is out, just write.
1 of 4