Commitments
Data & ownership pledge
Your workspace is yours — the documents, apps, prompts and knowledge you build up over time. This pledge is how we make sure it stays that way.
Last updated October 15, 2026
What we promise
- We don't train models on your work. sosofresh does not use the content of your workspace to train AI models — ours or anyone else's.
- We don't sell your data. Not to advertisers, not to data brokers, not to anyone.
- You can take it with you. Your work is stored in formats you can export, and leaving alexandr should never mean leaving your work behind.
- You choose where it lives. On your own computer, in our cloud, or inside your own infrastructure. The choice is yours, and it doesn't cost you features.
On your computer, it stays on your computer
When you run alexandr on your own Mac or PC, your workspace lives on that machine. The runtime is built into the desktop app. We never receive a copy of your documents or apps. The only data that leaves your machine is what's needed for the task you asked for — such as the conversation you send to the AI model you chose.
In our cloud, it stays isolated
Every cloud workspace runs in its own isolated runtime, with its own storage. One workspace can't see another's files, processes or database. We access a workspace only when it's necessary to operate the service, keep it secure, or when you ask us to help.
About the model providers
When you use a model from Anthropic, OpenAI, Google, Mistral or another provider, the conversation you send is processed by that provider under its own terms. We choose providers and settings that don't train on your requests wherever that option exists, and we're open about which providers your data goes to. If you want nothing to leave your machine at all, alexandr works with local models too.
If this ever changes
If we ever need to change this pledge, we'll announce it here and in our news before the change takes effect — and any change will apply only to data going forward, never to what you've already entrusted to us.
More commitments
alexandr's principles
How the assistant inside alexandr is meant to think and behave — and what it should never do.
Model neutrality pledge
A public promise to keep alexandr open to every model provider — and never lock you into one.
Transparency & trust
What we publish about how alexandr is built, run and secured, and how we hold ourselves to it.