Adam AIDocumentation adamai.co.ukSign in
Documentation › The layers
Open

People layer

Roles, Markings, approvals and the human checkpoint. Who may see and do what.

The People layer answers one question: who may see and do what. Everything else in Adam AI is built on the assumption that this layer's answer is correct and enforced, so it is the first layer and the one we are strictest about.

What it holds

The seeded Markings

Every Adam AI box starts with the same seven Markings so that nothing sensitive is unmarked on day one.

MarkingCoversDefault holders
PIIPersonal data about identifiable peopleLeadership, admin, named handlers
Special categoryHealth, beliefs, biometrics and other special category dataLeadership, named handlers
FinancialAccounts, payroll, banking, pricingFinance, leadership
Legal privilegeAdvice and correspondence under privilegeLegal, leadership
BoardBoard papers and minutesBoard, leadership
Client confidentialMaterial a customer has shared in confidenceThe customer's own team
CredentialsKeys, passwords, tokensNo agent may hold this, ever

Owners add their own: a Marking per client, per project, per product line. Managers of a Marking decide who holds it; the platform enforces the decision everywhere.

What it never does

Why it matters

We hear the same story often: a company told everyone not to share data with AI, and an audit then showed most staff, including most of the board, in breach of their own policy. People will use AI. The People layer gives them a safer way that is also the easier way, so they stop building their own.

Related: Security layer, Markings for your business.