Soriela reads the systems your projects live in, so where it runs and what it keeps are fair questions to ask first. This page answers them, and marks each deployment option with how it is actually offered today rather than implying they are all a switch away.
What changes between these is where the software runs, which models it calls, and how much of the operational burden sits with you. The status next to each one is how it is offered today — check it before you plan around it.
The fastest start. Shared infrastructure, with each workspace separated from every other.
You connect your sources and start the same day. We run and patch the infrastructure. This is where almost every team begins, and it is the option we run ourselves.
A dedicated environment for your organisation, set up with us rather than self-serve.
The same managed service, with your workspace on its own database and its own compute rather than sharing them. Usually the answer when an internal policy rules out shared infrastructure. Talk to us before you plan around it — we will tell you the lead time and what it involves.
Soriela deployed inside your own infrastructure, pointed at the model endpoint you choose.
Runs in your cloud or data centre, calling the models you already approve. Scoped with your team before deployment: we work through the infrastructure, model access and support arrangement together first.
Scoped with us for environments that allow no outbound connectivity.
Air-gapped environments are assessed case by case: what the architecture would have to look like, which models could run inside the boundary, and how it would be supported. We have not shipped one off the shelf, so treat this as a conversation to start early rather than a configuration to switch on.
If your security team has questions, put them in the room. That's our favourite kind of demo — and it's faster than a document round trip.
You choose the deployment, the connections and the boundary. Here is what happens to the project data you connect.
Soriela reads only the sources you connect, with only the permissions you approve when you connect them. Disconnect any source and it stops reading it immediately. Nothing is discovered or crawled on its own.
AI-generated insights — risks, actions and drafted updates — are grounded in the project sources they were read from and link back to them, so you can check why something was flagged instead of taking it on faith.
If shared infrastructure is not an option for you, Soriela can be deployed into your own environment on models you choose. Those options are set up with us rather than self-serve.
Your project data powers your own workspace. It is not used to train shared or third-party models, and we do not resell it.
Only the sources you connect, and only the scopes you approve when you connect them. Each source is assigned a role — plan, RAID log, tracker, chat, mailbox or meeting notes — so Soriela knows how to read it. Nothing is discovered or crawled on its own.
We don't train shared or third-party models on your project data. We don't resell it, and we don't hand it to advertisers or data brokers. Your data powers your workspace and nothing else.
Disconnect any source and Soriela stops reading it immediately. Close the workspace and it stops drawing data altogether. Ask us to delete it and we will.
Workspaces are separate from one another, each with its own projects, sources and dashboards. People are invited by email and hold an admin or a member role, and an activity feed records what changed.
30 minutes · live with the founders · see Soriela working, not a slide deck
