When we say the product is browser-resident, we mean something specific and load-bearing, not a marketing phrase. It is the architecture the whole product is built on. This entry sets out what it means and why it changes the security question entirely.
Where your matters live
Your matter data lives in your browser's own local storage, on your own device. Client names, matter facts, documents you have opened, the drafts you have produced — they sit on your disk, held by the browser the way a desktop application holds its files.
They do not sit in a database we run. There is no server of ours with a copy of your files on it. This is not a setting you switch on or a premium tier you upgrade to. It is how the product is built, from the first module to the last.
The security question this answers
Most conversations about AI security get stuck on how well a vendor protects the data you send them — the encryption, the certifications, the retention policy. All of that assumes the data has already left your machine and landed on theirs.
Browser-resident architecture removes that assumption. If there is no vendor database of your matters, there is:
- Nothing to breach. An attacker who gets into our systems does not find your files, because your files were never there.
- Nothing to subpoena. A third party cannot compel from us a copy of your matter that we do not hold.
- Nothing to forget to delete. There is no retention policy to trust, because there is no retained copy to depend on it.
The strongest protection for a privileged file is for it never to leave your control in the first place. That is what this design gives you.
What leaves, and what never does
Being browser-resident does not mean the product cannot use an AI model — it means it uses one on your terms. So it is worth being exact about the two categories.
What never leaves: your matter data at rest. The files on your disk, the full contents of a document, everything you have not specifically approved for a step. None of it is uploaded in the background. None of it is synced to a cloud we control. It stays put.
What leaves, and only then: the specific text you approve for one step. When you ask for a draft or an analysis, the assistant needs some text to work from, and that text goes to an AI model. But it is only the text that step requires, it passes the privilege gate in your browser first, and you see exactly what it is before it goes. Nothing travels that you have not seen and approved.
That is the whole boundary. Everything on one side stays on your machine. The only thing that crosses is what you decide to send, one step at a time, with a human on the decision every time.
Why we built it this way
We could have taken the easier road — a server, a database, your files uploaded so the product had everything to hand. It is simpler to build. It is also the exact shape a careful firm should refuse, because it hands a privileged file to a party you do not control before the work even begins.
Browser-resident was harder to build and it is the reason the product can honestly answer the question every solicitor should ask: where does my matter actually live? On your machine. It never left.
← All posts