Sense · Trust

AI you can put in front of an auditor.

Permissions, approvals, budgets and audit — the rules every Sense surface obeys, from the assistant to your own custom agents.

SEES

Only what you can see

Every answer is scoped to the asker’s record access — field-level permissions included.

WRITES

Only with approval

Every write is a draft until a human says yes — reassignments, reminders, reports, all of it.

ACCOUNTS

For every action

Questions, answers, approvals, and actions land in the same audit trail as everything else.

01 · The guarantees

Six rules that never bend.

Runs as you

Runs-as-you identity on every tool call — row- and field-level permissions inherited, every time.

Approval-gated writes

Configuration can add approvals, never remove them — the approval-required set is union-only.

Budgets before the call

Credits and quotas are checked before the model call, with daily token and cost budgets — and bring-your-own-key if you’d rather meter it yourself.

Knows what it’s not for

Off-topic questions are declined; admin questions are redirected to the right surface.

Model governance

Platform models are read-only per tenant; tenant models are BYO-key, encrypted.

One audit trail

Questions, answers, approvals and actions land in the same audit trail as everything else.