Security
Athena and your data
What she sees, what she never does, and how to switch her off.
Athena thinks with Claude, made by Anthropic: Haiku 5.5, Sonnet 5.5, Opus 5.5. The quickest one sorts and checks; the bigger ones plan, write and work through hard problems.
What is true today
Only your own organisation’s records
Athena reads only your workspace, and only what the person she is working for may see. The database itself keeps every other organisation out.
What only a person does
Issuing a permit, signing or removing an isolation, closing an incident and approving a payment are done by people. Athena has no way to do any of them, at any setting.
You set how far she goes
Each of Athena’s jobs has a level, from only watching to acting and telling you. Anything that touches safety stops at asking you first, whatever the level says.
Every decision shows why
Each thing Athena does or suggests is recorded with what she looked at and her reason, and you can see it from the record it changed.
Switch her off
An Owner or Admin can pause Athena for the whole organisation, or one job, at any time. Everything else carries on without her.
Words from outside are only words
Text that arrives in emails, documents, photos and forms is read as information, never as instructions to Athena.
We train nothing on your data
We do not train or fine-tune any model on your records.
Priced in credits, never by surprise
Heavy AI work costs credits, shown before it starts. Safety messages never cost credits and never stop.
Being confirmed with Anthropic
We say these only once we have them in writing from Anthropic, with the date we checked. Until then we make no claim about them:
- Whether the AI provider trains on it
- How long the AI provider keeps it
- Where the AI runs
- How long we keep what she looked at