Your data in a vault you actually own.
Your data. Your brain. Your terms.
Datta is a personal vault for the data that describes you, and an AI brain that reads it only when you say so. Every fact carries where it came from. Everything can be taken out. Nothing locks you in.
One email when Datta opens. No spam.
The Vault
Everything that describes you, each item traced to where it came from.
The Brain
Reads the vault, remembers only what you allow.
The agents
Get access that is scoped, time bound, and revocable. Nothing more.
The problem
Your life is already data. Almost none of it is yours to move.
A bank holds a fragment. A clinic holds another. An airline, an employer, a government portal, a dozen apps you forgot you signed up for. Each one keeps its own copy, sets its own rules, and hands you a download button if you are lucky. Now AI agents want that same context, and the fastest way to give it to them is to hand everything over one more time.
Scattered by default
Every service holds a piece of you and files it as their record. You never see the whole, and you cannot move it.
Impossible to trust
A file in a folder says nothing about where it came from. An agent acting on it is guessing, and so is everyone downstream.
Borrowed, never owned
The moment you paste your context into someone else's model, you have given away the one thing that was still yours.
What Datta is
Two halves of one idea: a vault you own, and a brain that answers to you.
Everything that describes you, in one place you control.
Documents, identifiers, records, the facts that add up to a person or a company. Each item is stored with its origin and how far it has been verified, so what comes out is traceable instead of assumed.
- You decide what goes in
- Each item keeps its origin and verification level
- Export the whole thing, whenever you want
- Close the account and it leaves with you
Your context and memory for AI, kept on your side of the line.
The brain is what an agent reads when it needs to know something about you. It works from the vault, remembers only what you decide it should, and answers to the permissions you set rather than to whoever built the model.
- Reads the vault only with your permission
- Access is scoped and time bound
- Revoke any agent, at any moment
- Your data is never training data
How it works
Four moves. That is the whole model.
Put it in
Bring in documents and data, from a photo of an ID to a record pulled from a service you already use.
It gets traced
Each item is stored with where it came from and how strongly it has been verified. Nothing arrives anonymous.
Grant access
An agent, an app, or a company asks for something specific. You grant that, scoped, and for as long as you choose.
Take it back
Revoke a permission, export everything, or close the account and walk out with the whole vault.
The premise
Ownership has to be built, not promised.
Every company says it respects your data. The real question is whether you can walk out with it. These are the constraints Datta is built under, and the ones we intend to be held to.
You own it, we hold it
Your data is not licensed to us and it is not an asset on our books. We are a custodian you can fire.
Consent is explicit and revocable
Nothing is shared by default. Every grant has a scope and an end date, and you can end it earlier.
Export is a feature, not a favor
Taking everything out and deleting your account are part of the account itself, from day one, not a support ticket.
No lock-in, on any layer
Datta runs on infrastructure we can pick up and move. No managed cloud service sits in the critical path, because a company that is locked in cannot credibly promise that you are not.
Who it is for
For people
Your identity, your records, your history, held in one place and lent out on your terms.
For companies
In developmentThe same model at company scale: its documents, its people, its verified facts, ready for the agents it chooses to authorize.
Early access
Be first when Datta opens.
Join the list. We will write the moment it opens, and the people who join early help shape what it becomes.
Questions
The things people ask first.
When does Datta open?
Early access first, in waves, so we can onboard people properly. We are not putting a date on the public launch until the first wave is running. Join the list and you will hear from us before anyone else.
Where does my data actually live?
On servers we operate directly, today in the United States, not on a managed cloud service that holds the keys for us. The stack is deliberately portable, so where it runs can change without the product changing. Which regions come next is one of the things early members help decide.
Can I really get everything out?
Yes. Exporting your data and deleting your account are built into the account itself, not a request you file with support. If leaving is hard, ownership was never real.
Do you train AI on my data?
No. Your vault is not training data, not for us and not for anyone we work with. The brain works for you, on your data, under permissions you set.
What is an AI brain, concretely?
It is the layer that lets an agent know things about you without you pasting your life into a chat window. It reads from the vault, only what you allowed, only for as long as you allowed it.
What will it cost?
Pricing is not set yet, and we would rather say so than invent a number. Early access members will know before anyone else, and they will not be surprised by it.