SECURITY
Built to be questioned. By construction, not by feature.
Most systems are built to produce an answer. This one is built for the conversation a year later, when someone asks where the answer came from. Security here is not a layer added on top — it is a property of how the platform is arranged.
PRINCIPLES
Four commitments the architecture is arranged to keep.
These are not features. They are properties of the platform — true before the first customer signs and after.
Traceable
Every claim resolves to the evidence under it. Nothing stands on its own assertion — a decision can be followed back to what it rested on at the time it was made.
Explainable
A decision is stored with the reasoning it had at the time, not the reasoning reconstructed for the review. What you can explain tomorrow is what you actually decided today.
Human-reviewed
Evidence is accepted by a person. The system proposes; it does not admit its own inputs. No evidence enters the record without a human decision behind it.
Governed
Consent, access, retention and audit are properties of the platform, not switches somebody has to remember. They apply because of how the system is built, not because of how it is configured.
ARCHITECTURE
How the platform enforces them.
The Trust Architecture is the mechanism behind the four commitments. It is described here at the level a technical evaluator needs; the implementation detail lives in the codebase.
Consent as a precondition
Access to a person's evidence requires their consent, recorded and revocable. The system does not have a path to data that bypasses it.
Role-based access
What a user can see and do is determined by their role within an organization, enforced at the service layer rather than hidden in the UI.
Retention by design
How long evidence is kept is a property of the record type, not a manual setting. The system ages records according to rules that were decided before the data existed.
Audit by default
Decisions are recorded with their reasoning at the time they were made. The audit trail is not assembled after the fact — it is a byproduct of how decisions are produced.
DATA HANDLING
What we do — and do not do — with your data.
Stated plainly, because the things a platform does not do with your data matter as much as the things it does.
Your private content is not training data
Evidence, assessments, decisions and personal content are never used to train public models. The boundary between your data and model training is structural, not procedural.
Encryption in transit and at rest
Data is encrypted while moving between you and OZLILI, and while stored. This is baseline, not a differentiator — stated here because omitting it would be worse than stating it.
You can see what we hold
The evidence and decisions on record are inspectable by the person they concern. Transparency is a property of the system, not a request you have to make.
Have a security question?
For security reviews, due diligence, or specific technical questions, contact the team directly. We answer detailed questions from evaluators — this is not a marketing redirect.