Trust
What we can prove, and what we cannot.
No customer logos, no certifications and no case studies — because there are none to show. What follows is the actual state of the company and of your data in it.
Where we are
The questions a security review asks, answered before you ask them.
This is an early-stage product from a small, unfunded team. Every claim on this site is meant to carry its evidence, and that has to include the claims that are not flattering.
How far along is this?
Early. The product is live and in use, and it is not a mature platform with years of production behaviour behind it. Deployments today are close-contact: you will be talking to the people who build it.
What certifications do you hold?
None. No SOC 2, no ISO 27001, no HIPAA or FedRAMP attestation. The SOC 2 sign-off visible in the product is a customer recording their own attestation against their own standard — it is a feature, not a certification we hold.
Who is behind it, and who funds it?
Limina Labs. No outside investment. That means no board pressure on the roadmap, and it also means a small team — worth weighing honestly against a vendor with a support organisation behind it.
Is the source available?
No. The repositories are private, and running it inside your own network is under a commercial agreement rather than a public licence. If your procurement requires source escrow, that is a conversation, not a blocker.
What happens if you stop?
The record is yours and it is portable — intents, evidence and published bundles export as files you already know how to read. A self-hosted deployment keeps running on infrastructure you control. Ask about escrow terms before you sign anything.
What should stop us buying today?
A hard requirement for a certification we do not hold, a procurement process that will not clear an unfunded vendor, or a need for 24/7 supported operations. Those are real, and none of them get better by us being vague about them.
Your data
What we hold, and what never leaves.
Two deployment shapes with genuinely different answers. Which one you need is a procurement question, and it is worth settling early.
What leaves your network
On a self-hosted deployment, the model provider you configure is the only runtime external dependency. Nothing else calls out. On the managed service, your intents and evidence sit in our infrastructure.
What we keep
Intents, the evidence attached to them, and an append-only record of who decided what and when. Deletion is a state rather than an erasure, so a past decision stays reconstructable — which is the point of the record, and worth knowing before you put anything sensitive in it.
What we do not do
We do not train models on your content. We do not sell or share it. There is no download button anywhere in the product: hand-off is a scoped, revocable credential bound to a published version, with a read log, because a downloaded file could not be recalled.
Who can see what
Membership, permission and accountability are separate. A guest is scoped to one intent and never the workspace, and an outside auditor can attest without ever seeing it.
The full deployment posture — managed service or your own infrastructure, from the same artifact — is on the Autonomy page.
See it on one of your own intents.
The fastest way to understand Intent Studio is to walk something you are actually working on through it. Half an hour, your intent, no slides.
A demo is a conversation with the people who build it. Try it now opens the app sign-in.